Pricing · verified Nov 2, 2026 · 58 models · 6 providers · history since Sep 2021
LLM API pricing, with the history nobody else keeps
Current per-million-token prices for every model PolyCog offers — every cut, increase and promotion since launch, and every retired model back to each provider’s first API price, dated and sourced. PolyCog itself is free, $10 a month or $250 once; the AI is billed by the providers to your own keys at exactly these prices.
ChatGPT
flagship$10.00 / $50.00
cheapest$0.10 / $0.50
▼ GPT-5.6 Sol · Aug 21, 2026
Claude
flagship$10.00 / $50.00
cheapest$0.10 / $0.50
no changes on record
Gemini
flagship$2.00 / $10.00
cheapest$0.10 / $0.40
▼ Gemini 3.6 Flash · Aug 13, 2026
DeepSeek
flagship$1.32 / $3.96
cheapest$0.30 / $1.20
▲ DeepSeek V4 Pro · Aug 16, 2026
Grok
flagship$2.00 / $6.00
cheapest$1.25 / $2.50
no changes on record
Wildcard
flagship$2.00 / $6.00
cheapest$0.15 / $0.47
▼ Mistral Large 4 · ≈ Nov 2, 2026
How prices moved across all 6 providers
Each provider’s flagship and cheapest text model on one log scale, back to Sep 2021. The Wildcard spans eight labs, so its top line is the priciest model it offers. Tap a provider to hide or show it, and hover or drag across the chart to read any date.
Lines
Range
Show
FlagshipCheapest
same model repriced
a new model took the slot
blended 3 in : 1 out · USD per 1M tokens · log scale
Flagship prices over the last 12 months. Every provider’s own chart below goes back to its launch.
Current prices by provider
USD per one million tokens, standard tier, provider list price. Tags mirror the app: synthesis = the bot’s flagship pick, default = its everyday model, sample = available on the keyless preview, more = behind the picker’s expander.
▲ redprice went up (or will)
▼ greenprice went down
greyunchanged
≈date approximate
USD per 1M tokens · standard tier · provider list price
ChatGPT · OpenAI API
Verified Nov 2, 2026 against the OpenAI pricing pageSince July: 8 new models (GPT-6 Astra, GPT-6.1 Sol, GPT-6.1 Terra…) · 3 price cuts (GPT-5.6 Sol, GPT-5.6 Terra, GPT-5.6 Luna) · 1 announced future change.
| Model | Input / 1M | Output / 1M | Blended 3:1 | Since launch | Last change |
|---|---|---|---|---|---|
| GPT-6 Astra synthesis profile → gpt-6-astra | $10.00 | $50.00 | $20.00 | unchangedsince Sep 2026 | — |
| GPT-6.1 Sol profile → gpt-6.1-sol | $2.00 | $10.00 | $4.00 | unchangedsince Sep 2026 | — |
| GPT-6.1 Terra profile → gpt-6.1-terra | $1.00 | $5.00 | $2.00 | unchangedsince Oct 2026 | new · Oct 26, 2026 |
| GPT-6 Luna default profile → gpt-6-luna | $0.10 | $0.50 | $0.20 | unchangedsince Sep 2026 | — |
| GPT-5.6 Sol more promo → Nov 21, 2026 profile → gpt-5.6-sol ▲ $5.00 / $30.00 after Nov 21, 2026 if the promo lapses | $4.00 | $20.00 | $8.00 | ▼ 20% in▼ 33% outsince Jul 2026 | ▼ Aug 21, 2026 |
| GPT-6 Sol more profile → gpt-6-sol | $2.00 | $10.00 | $4.00 | unchangedsince Sep 2026 | — |
| GPT-5.6 Terra more profile → gpt-5.6-terra | $2.00 | $12.00 | $4.50 | ▼ 20% in▼ 20% outsince Jul 2026 | ▼ Jul 30, 2026 |
| GPT-5.6 Luna more profile → gpt-5.6-luna | $0.20 | $1.20 | $0.45 | ▼ 80% in▼ 80% outsince Jul 2026 | ▼ Jul 30, 2026 |
| GPT-5.5 more profile → gpt-5.5 | $5.00 | $30.00 | $11.25 | unchangedsince Apr 2026 | — |
| GPT-5.4 more profile → gpt-5.4 | $2.50 | $15.00 | $5.625 | unchangedsince Mar 2026 | — |
| GPT-5.1 more profile → gpt-5.1 | $1.25 | $10.00 | $3.438 | unchangedsince Nov 2025 | — |
| GPT-4.1 more profile → gpt-4.1 | $2.00 | $8.00 | $3.50 | unchangedsince Apr 2025 | — |
| GPT-5.4 mini more profile → gpt-5.4-mini | $0.75 | $4.50 | $1.688 | unchangedsince Mar 2026 | — |
| GPT-4.1 mini more profile → gpt-4.1-mini | $0.40 | $1.60 | $0.70 | unchangedsince Apr 2025 | — |
| GPT-4o mini more profile → gpt-4o-mini | $0.15 | $0.60 | $0.263 | unchangedsince Jul 2024 | — |
| No models match the current filters. | |||||
ChatGPT — price history
Nov 22, 2026
GPT-5.6 SolPossible increaseif promo lapsesin$4.00$5.00▲ 25%out$20.00$30.00▲ 50%
Only if the promo lapses — OpenAI has not announced an end price; the pre-promo list was $5 / $30.
Oct 26, 2026
GPT-6.1 TerraLaunchedin$1.00out$5.00
Cached input $0.05. Cache writes $1.25 (30-minute) / $2.50 (12-hour). Prompts over 272K tokens bill 2× input and cache, 1.5× output ($2 / $7.50). OpenAI: “Standard API pricing is $1 per million input tokens, $0.05 per million cached input tokens, and $5 per million output tokens.” The first Terra since GPT-5.6 Terra ($2 / $12). source
Sep 29, 2026
GPT-6.1 SolLaunchedin$2.00out$10.00
Cached input $0.10 (5% of input; GPT-6 Sol's is 10%). Cache writes $2.50. Prompts over 272K tokens bill $4 / $15. OpenAI: “Near-Astra performance for complex work at a lower cost.” source
Sep 22, 2026
GPT-6 LunaLaunchedin$0.10out$0.50
Cached input $0.01. Prompts over 272K tokens bill $0.20 / $0.75. source
Sep 22, 2026
GPT-6 SolLaunchedin$2.00out$10.00
Cached input $0.20. Prompts over 272K tokens bill $4 / $15. OpenAI: "50% cheaper" than GPT-5.6 Sol's promotional price. source
Sep 3, 2026
GPT-6 AstraLaunchedin$10.00out$50.00
Cached input $1.00. Prompts over 272K tokens bill 2× input / 1.5× output. source
Sep 2, 2026
GPT-5.6 SolPolyCog’s estimate
PolyCog’s estimate stayed at $5 / $30 from Aug 21 until Sep 2, 2026 (over-estimated 20–33%).
Sep 2, 2026
GPT-5.6 TerraPolyCog’s estimate
PolyCog’s estimate stayed at $2.50 / $15 from Jul 30 until Sep 2, 2026 (over-estimated 25%).
Sep 2, 2026
GPT-5.6 LunaPolyCog’s estimate
PolyCog’s estimate stayed at $1 / $6 from Jul 30 until Sep 2, 2026 (over-estimated 5×).
Aug 21, 2026
GPT-5.6 SolPromotional priceuntil Nov 21, 2026in$5.00$4.00▼ 20%out$30.00$20.00▼ 33%
OpenAI: “promotional pricing is available at least through November 21, 2026.” source
Jul 30, 2026
GPT-5.6 TerraPrice cutin$2.50$2.00▼ 20%out$15.00$12.00▼ 20%
OpenAI: “Starting July 30, GPT-5.6 Luna costs 80% less, while GPT-5.6 Terra costs 20% less.” source
Jul 30, 2026
GPT-5.6 LunaPrice cutin$1.00$0.20▼ 80%out$6.00$1.20▼ 80%
An 80% permanent cut, announced with Terra’s. source
Nov 13, 2025
GPT-5.1Launchedin$1.25out$10.00
Deprecated 2026-10-01; OpenAI shutdown 2027-04-01 (replacement: gpt-6-sol). source
Jun 10, 2025
o3Price cutretires Dec 2026in$10.00$2.00▼ 80%out$40.00$8.00▼ 80%
80% price cut across the board; OpenAI framed it as passing through efficiency gains, alongside the launch of o3-pro. source
Feb 27, 2025
GPT-4.5 (research preview)Launchedretired Jul 2025secondary sourcein$75.00out$150.00
Aug 6, 2024
GPT-4oPrice cutretires Jan 2027secondary sourcein$5.00$2.50▼ 50%out$15.00$10.00▼ 33%
gpt-4o-2024-08-06 snapshot: 50% input cut, 33% output cut; output-token limit quadrupled same day. source
Jan 25, 2024
GPT-3.5 TurboPrice cutretired Oct 2026in$1.00$0.50▼ 50%out$2.00$1.50▼ 25%
gpt-3.5-turbo-0125 released: input cut 50%, output cut 25%. Became OpenAI's cheapest surviving model after the Jan 4 2024 GPT-3 shutdown left a brief gap. source
Nov 6, 2023
GPT-3.5 TurboPrice cutretired Oct 2026in$1.50$1.00▼ 33%out$2.00
DevDay: gpt-3.5-turbo-1106 (16K context) cut input 3x vs the old 16K model. source
Nov 6, 2023
GPT-4 TurboLaunchedretired Oct 2026in$10.00out$30.00
Originally priced as $0.01 per 1K input tokens / $0.03 per 1K output tokens. source
Jun 13, 2023
GPT-3.5 TurboPrice cutretired Oct 2026in$2.00$1.50▼ 25%out$2.00
Split into separate input/output pricing for the first time (25% input cut); gpt-3.5-turbo-0613 and a 16K-context variant (gpt-3.5-turbo-16k, $3/$4 per 1M) also introduced same day. source
Mar 14, 2023
GPT-4 (8K)Launchedretired Oct 2026in$30.00out$60.00
Originally priced as $0.03 per 1K prompt tokens / $0.06 per 1K completion tokens. source
Mar 14, 2023
GPT-4 (32K)Launchedretired Jun 2025in$60.00out$120.00
Originally priced as $0.06 per 1K prompt tokens / $0.12 per 1K completion tokens. source
Mar 1, 2023
GPT-3.5 TurboLaunchedretired Oct 2026in$2.00out$2.00
Originally priced as $0.002 per 1K tokens (single blended rate). source
Nov 28, 2022
text-davinci-003Launchedretired Jan 2024secondary sourcein$20.00out$20.00
Originally priced as $0.0200 per 1K tokens (single rate). source
Sep 1, 2022
GPT-3 DavinciPrice cutretired Jan 2024secondary sourcein$60.00$20.00▼ 67%out$60.00$20.00▼ 67%
Two-thirds price cut across the GPT-3 engine family, effective Sept 1 2022. source
Sep 1, 2022
GPT-3 CuriePrice cutretired Jan 2024secondary sourcein$6.00$2.00▼ 67%out$6.00$2.00▼ 67%
Two-thirds price cut across the GPT-3 engine family. source
Sep 1, 2022
GPT-3 BabbagePrice cutretired Jan 2024secondary sourcein$1.20$0.50▼ 58%out$1.20$0.50▼ 58%
Two-thirds-class price cut across the GPT-3 engine family. source
Sep 1, 2022
GPT-3 AdaPrice cutretired Jan 2024secondary sourcein$0.80$0.40▼ 50%out$0.80$0.40▼ 50%
Two-thirds-class price cut across the GPT-3 engine family. source
≈Sep 1, 2021
GPT-3 DavinciLaunchedretired Jan 2024in$60.00out$60.00
Originally priced as $0.0600 per 1K tokens (single rate). source
≈Sep 1, 2021
GPT-3 CurieLaunchedretired Jan 2024in$6.00out$6.00
Originally priced as $0.0060 per 1K tokens (single rate). source
≈Sep 1, 2021
GPT-3 BabbageLaunchedretired Jan 2024in$1.20out$1.20
Originally priced as $0.0012 per 1K tokens (single rate). source
≈Sep 1, 2021
GPT-3 AdaLaunchedretired Jan 2024in$0.80out$0.80
Originally priced as $0.0008 per 1K tokens (single rate). source
No dated history for this selection yet.
Since launch
OpenAI API since Jun 2020 · first public price Sep 2021Flagship: GPT-3 Davinci at $60.00 / $60.00 → GPT-6 Astra at $10.00 / $50.00 today. Cheapest: GPT-3 Ada at $0.80 / $0.80 → GPT-5 nano at $0.05 / $0.40.
Flagship — the provider’s best model at the timeCheapest text modelblended 3 in : 1 out · USD per 1M tokens · log scale
Hover or tap for the exact input / output at any change · ← → on a focused chart · 22 dated changes
All 40 OpenAI models since Sep 2021 — launch and last price
| Model | Tier | Launched | Launch in / out | Last in / out | Retired |
|---|---|---|---|---|---|
| GPT-3 Davinci≈ datedavinci≈ Sep 1, 2021 | flagship | ≈ Sep 1, 2021 | $60.00 / $60.00 | $20.00 / $20.00▼ 67% in ▼ 67% out | Jan 4, 2024 |
| GPT-3 Curie≈ datecurie≈ Sep 1, 2021 | mid | ≈ Sep 1, 2021 | $6.00 / $6.00 | $2.00 / $2.00▼ 67% in ▼ 67% out | Jan 4, 2024 |
| GPT-3 Babbage≈ datebabbage≈ Sep 1, 2021 | small | ≈ Sep 1, 2021 | $1.20 / $1.20 | $0.50 / $0.50▼ 58% in ▼ 58% out | Jan 4, 2024 |
| GPT-3 Ada≈ dateada≈ Sep 1, 2021 | small | ≈ Sep 1, 2021 | $0.80 / $0.80 | $0.40 / $0.40▼ 50% in ▼ 50% out | Jan 4, 2024 |
| text-davinci-003secondarytext-davinci-003Nov 28, 2022 | flagship | Nov 28, 2022 | $20.00 / $20.00 | $20.00 / $20.00 | Jan 4, 2024 |
| GPT-3.5 Turbogpt-3.5-turboMar 1, 2023 | flagship | Mar 1, 2023 | $2.00 / $2.00 | $0.50 / $1.50▼ 75% in ▼ 25% out | Oct 23, 2026 |
| GPT-4 (8K)gpt-4Mar 14, 2023 | flagship | Mar 14, 2023 | $30.00 / $60.00 | $30.00 / $60.00 | Oct 23, 2026 |
| GPT-4 (32K)gpt-4-32kMar 14, 2023 | flagship | Mar 14, 2023 | $60.00 / $120.00 | $60.00 / $120.00 | Jun 6, 2025 |
| GPT-4 Turbogpt-4-turboNov 6, 2023 | flagship | Nov 6, 2023 | $10.00 / $30.00 | $10.00 / $30.00 | Oct 23, 2026 |
| GPT-4onot offeredsecondaryprofile →gpt-4oMay 13, 2024 | flagship | May 13, 2024 | $5.00 / $15.00 | $2.50 / $10.00▼ 50% in ▼ 33% out | retires Jan 14, 2027 |
| GPT-4o miniofferedprofile →gpt-4o-miniJul 18, 2024 | small | Jul 18, 2024 | $0.15 / $0.60 | $0.15 / $0.60 | — |
| o1-previewsecondaryo1-previewSep 12, 2024 | reasoning | Sep 12, 2024 | $15.00 / $60.00 | $15.00 / $60.00 | Jul 28, 2025 |
| o1-minisecondaryo1-miniSep 12, 2024 | reasoning | Sep 12, 2024 | $3.00 / $12.00 | $3.00 / $12.00 | Oct 27, 2025 |
| o1secondaryo1Dec 17, 2024 | reasoning | Dec 17, 2024 | $15.00 / $60.00 | $15.00 / $60.00 | Oct 23, 2026 |
| o3-minisecondaryprofile →o3-miniJan 31, 2025 | reasoning | Jan 31, 2025 | $1.10 / $4.40 | $1.10 / $4.40 | Oct 23, 2026 |
| GPT-4.5 (research preview)secondarygpt-4.5-previewFeb 27, 2025 | other | Feb 27, 2025 | $75.00 / $150.00 | $75.00 / $150.00 | Jul 14, 2025 |
| GPT-4.1offeredprofile →gpt-4.1Apr 14, 2025 | mid | Apr 14, 2025 | $2.00 / $8.00 | $2.00 / $8.00 | — |
| GPT-4.1 miniofferedprofile →gpt-4.1-miniApr 14, 2025 | small | Apr 14, 2025 | $0.40 / $1.60 | $0.40 / $1.60 | — |
| GPT-4.1 nanonot offeredgpt-4.1-nanoApr 14, 2025 | small | Apr 14, 2025 | $0.10 / $0.40 | $0.10 / $0.40 | still served |
| o3not offeredsecondaryprofile →o3Apr 16, 2025 | reasoning | Apr 16, 2025 | $10.00 / $40.00 | $2.00 / $8.00▼ 80% in ▼ 80% out | retires Dec 11, 2026 |
| o4-minisecondaryo4-miniApr 16, 2025 | reasoning | Apr 16, 2025 | $1.10 / $4.40 | $1.10 / $4.40 | Oct 23, 2026 |
| GPT-5not offeredgpt-5Aug 7, 2025 | flagship | Aug 7, 2025 | $1.25 / $10.00 | $1.25 / $10.00 | retires Dec 11, 2026 |
| GPT-5 mininot offeredgpt-5-miniAug 7, 2025 | mid | Aug 7, 2025 | $0.25 / $2.00 | $0.25 / $2.00 | still served |
| GPT-5 nanonot offeredgpt-5-nanoAug 7, 2025 | small | Aug 7, 2025 | $0.05 / $0.40 | $0.05 / $0.40 | still served |
| GPT-5.1offeredprofile →gpt-5.1Nov 13, 2025 | flagship | Nov 13, 2025 | $1.25 / $10.00 | $1.25 / $10.00 | — |
| GPT-5.1 nanonot offeredsecondarygpt-5.1-nanoNov 13, 2025 | small | Nov 13, 2025 | $0.05 / $0.40 | $0.05 / $0.40 | still served |
| GPT-5.2not offeredgpt-5.2Dec 11, 2025 | flagship | Dec 11, 2025 | $1.75 / $14.00 | $1.75 / $14.00 | still served |
| GPT-5.3-Codexnot offered≈ dategpt-5.3-codex≈ Jan 1, 2026 | other | ≈ Jan 1, 2026 | $1.75 / $14.00 | $1.75 / $14.00 | still served |
| GPT-5.4offeredprofile →gpt-5.4Mar 5, 2026 | flagship | Mar 5, 2026 | $2.50 / $15.00 | $2.50 / $15.00 | — |
| GPT-5.4 miniofferedprofile →gpt-5.4-miniMar 17, 2026 | mid | Mar 17, 2026 | $0.75 / $4.50 | $0.75 / $4.50 | — |
| GPT-5.4 nanonot offeredgpt-5.4-nanoMar 17, 2026 | small | Mar 17, 2026 | $0.20 / $1.25 | $0.20 / $1.25 | still served |
| GPT-5.5offeredprofile →gpt-5.5Apr 24, 2026 | flagship | Apr 24, 2026 | $5.00 / $30.00 | $5.00 / $30.00 | — |
| GPT-5.6 Solofferedprofile →gpt-5.6-solJul 9, 2026 | flagship | Jul 9, 2026 | $5.00 / $30.00 | $4.00 / $20.00▼ 20% in ▼ 33% out | — |
| GPT-5.6 Terraofferedprofile →gpt-5.6-terraJul 9, 2026 | mid | Jul 9, 2026 | $2.50 / $15.00 | $2.00 / $12.00▼ 20% in ▼ 20% out | — |
| GPT-5.6 Lunaofferedprofile →gpt-5.6-lunaJul 9, 2026 | small | Jul 9, 2026 | $1.00 / $6.00 | $0.20 / $1.20▼ 80% in ▼ 80% out | — |
| GPT-6 Astraofferedprofile →gpt-6-astraSep 3, 2026 | flagship | Sep 3, 2026 | $10.00 / $50.00 | $10.00 / $50.00 | — |
| GPT-6 Lunaofferedprofile →gpt-6-lunaSep 22, 2026 | small | Sep 22, 2026 | $0.10 / $0.50 | $0.10 / $0.50 | — |
| GPT-6 Solofferedprofile →gpt-6-solSep 22, 2026 | mid | Sep 22, 2026 | $2.00 / $10.00 | $2.00 / $10.00 | — |
| GPT-6.1 Solofferedprofile →gpt-6.1-solSep 29, 2026 | mid | Sep 29, 2026 | $2.00 / $10.00 | $2.00 / $10.00 | — |
| GPT-6.1 Terraofferedprofile →gpt-6.1-terraOct 26, 2026 | mid | Oct 26, 2026 | $1.00 / $5.00 | $1.00 / $5.00 | — |
Claude · Anthropic API
Verified Nov 2, 2026 against the Claude pricing pageSince July: 5 new models (Claude Fable 5.1, Claude Opus 5.5, Claude Sonnet 5.5…).
| Model | Input / 1M | Output / 1M | Blended 3:1 | Since launch | Last change |
|---|---|---|---|---|---|
| Claude Fable 5.1 synthesis profile → claude-fable-5-1 | $10.00 | $50.00 | $20.00 | unchangedsince Sep 2026 | — |
| Claude Opus 5.5 profile → claude-opus-5-5 | $4.00 | $20.00 | $8.00 | unchangedsince Sep 2026 | — |
| Claude Sonnet 5.5 default profile → claude-sonnet-5-5 | $2.00 | $10.00 | $4.00 | unchangedsince Sep 2026 | — |
| Claude Haiku 5.5 sample profile → claude-haiku-5-5 | $0.10 | $0.50 | $0.20 | unchangedsince Oct 2026 | new · Oct 7, 2026 |
| Claude Haiku 4.5 more profile → claude-haiku-4-5-20251001 | $1.00 | $5.00 | $2.00 | unchangedsince Oct 2025 | — |
| Claude Sonnet 5 more profile → claude-sonnet-5 | $2.00 | $10.00 | $4.00 | unchangedsince Jun 2026 | — |
| Claude Fable 5 more profile → claude-fable-5 | $10.00 | $50.00 | $20.00 | unchangedsince Jun 2026 | — |
| Claude Opus 5 more profile → claude-opus-5 | $5.00 | $25.00 | $10.00 | unchangedsince Jul 2026 | — |
| Claude Opus 4.8 more profile → claude-opus-4-8 | $5.00 | $25.00 | $10.00 | unchangedsince May 2026 | — |
| Claude Opus 4.7 more profile → claude-opus-4-7 | $5.00 | $25.00 | $10.00 | unchangedsince Apr 2026 | — |
| Claude Opus 4.6 more profile → claude-opus-4-6 | $5.00 | $25.00 | $10.00 | unchangedsince Feb 2026 | — |
| Claude Sonnet 4.6 more profile → claude-sonnet-4-6 | $3.00 | $15.00 | $6.00 | unchangedsince Feb 2026 | — |
| No models match the current filters. | |||||
Claude — price history
Oct 7, 2026
Claude Haiku 5.5Launchedin$0.10out$0.50
Prompts up to 100K tokens; a prompt over 100K tokens pays $0.50 / $2.50 for the whole request (PolyCog models this tier). Cache read $0.01, 5-minute write $0.125. Uses the newer tokenizer (about 30% more tokens than Haiku 4.5 for the same text). Adaptive thinking on by default. source
Sep 28, 2026
Claude Sonnet 5.5Launchedin$2.00out$10.00
The same prices as Claude Sonnet 5: cache read $0.20; 5-minute cache write $2.50, 1-hour $4. Adaptive thinking on by default. source
Sep 22, 2026
Claude Opus 5.5Launchedin$4.00out$20.00
Cache read $0.20 (0.05×); 5-minute cache write $5, 1-hour $8. Fast mode (research preview) $8 / $40. Adaptive thinking always on. source
Sep 2, 2026
Claude Sonnet 5PolyCog’s estimate
PolyCog estimated at the announced $3 / $15 list price (a deliberate 1.5× over-estimate) from mid-July until Sep 2, 2026.
Sep 1, 2026
Claude Fable 5.1Launchedin$10.00out$50.00
Same rate card as Fable 5; cache reads cut to $0.25 per MTok. source
Aug 10, 2026
Claude Sonnet 5Introductory price made permanentin$2.00out$10.00
was to become $3.00 / $15.00 · Anthropic made the introductory price the standard price; the scheduled Sep 1 increase to $3 / $15 will not occur. source
≈Jul 21, 2026
Claude Haiku 4.5PolyCog’s estimate
PolyCog’s table carried $0.80 / $4.00 (a mis-entry) until mid-July 2026 — estimates were 20% low.
≈Jul 21, 2026
Claude Opus 4.6PolyCog’s estimate
PolyCog’s table carried $15 / $75 (a mis-entry) until mid-July 2026 — restored Opus 4.6 sessions were over-reported 3×.
Jun 30, 2026
Claude Sonnet 5Launched at an introductory priceuntil Aug 31, 2026in$2.00out$10.00
Launched at an introductory $2 / $10, scheduled to rise to $3 / $15 on Sep 1, 2026. source
Oct 15, 2025
Claude Haiku 4.5Launcheddate derivedin$1.00out$5.00
Eligible for retirement not sooner than Oct 15, 2026 (no notice issued). source
Dec 3, 2024
Claude 3.5 HaikuPrice cutretired Feb 2026in$1.00$0.80▼ 20%out$5.00$4.00▼ 20%
Cut from $1/$5 to $0.80/$4 (-20%) one month after launch; Anthropic appended a dated update to the October announcement post. source
≈Dec 13, 2023
Claude Instant (1.0 / 1.1 / 1.2)Price cutretired Nov 2024in$1.63$0.80▼ 51%out$5.51$2.40▼ 56%
Cut to $0.80/$2.40 (-51%/-56%). Date bracketed: Anthropic's Nov 2023 pricing PDF (which already lists Claude 2.1, so >= 2023-11-21) still shows Instant at $1.63/$5.51; the updated Anthropic PDF shows $0.80/$2.40 and a Hacker News post noticed the change on 202 source
Nov 21, 2023
Claude 2Price cutretired Jul 2025in$11.02$8.00▼ 27%out$32.68$24.00▼ 27%
Cut to $8/$24 (-27%) alongside the Claude 2.1 launch ("We are also updating our pricing to improve cost efficiency for our customers across models"). Anthropic's Nov 2023 pricing PDF lists Claude 2.0 and 2.1 both at $8.00/$24.00. source
≈Mar 14, 2023
Claude (v1: claude-1.0 / 1.1 / 1.2 / 1.3)Launchedretired Nov 2024in$11.02out$32.68
Originally priced as $11.02 / $32.68 per million tokens (Prompt / Completion) — figure verified from Anthropic's pricing PDF dated 2023-05-10. source
≈Mar 14, 2023
Claude Instant (1.0 / 1.1 / 1.2)Launchedretired Nov 2024in$1.63out$5.51
Originally priced as $1.63 / $5.51 per million tokens (Prompt / Completion) — verified from Anthropic's pricing PDF dated 2023-05-10; launch-. source
No dated history for this selection yet.
Since launch
Anthropic API since Mar 2023 · first public price Mar 2023Flagship: Claude (v1: claude-1.0 / 1.1 / 1.2 / 1.3) at $11.02 / $32.68 → Claude Fable 5.1 at $10.00 / $50.00 today. Cheapest: Claude Instant (1.0 / 1.1 / 1.2) at $1.63 / $5.51 → Claude Haiku 5.5 at $0.10 / $0.50.
Flagship — the provider’s best model at the timeCheapest text modelblended 3 in : 1 out · USD per 1M tokens · log scale
Hover or tap for the exact input / output at any change · ← → on a focused chart · 20 dated changes
All 28 Anthropic models since Mar 2023 — launch and last price
| Model | Tier | Launched | Launch in / out | Last in / out | Retired |
|---|---|---|---|---|---|
| Claude (v1: claude-1.0 / 1.1 / 1.2 / 1.3)≈ dateclaude-1≈ Mar 14, 2023 | flagship | ≈ Mar 14, 2023 | $11.02 / $32.68 | $11.02 / $32.68 | Nov 6, 2024 |
| Claude Instant (1.0 / 1.1 / 1.2)≈ dateclaude-instant-1≈ Mar 14, 2023 | small | ≈ Mar 14, 2023 | $1.63 / $5.51 | $0.80 / $2.40▼ 51% in ▼ 56% out | Nov 6, 2024 |
| Claude 2claude-2.0Jul 11, 2023 | flagship | Jul 11, 2023 | $11.02 / $32.68 | $8.00 / $24.00▼ 27% in ▼ 27% out | Jul 21, 2025 |
| Claude 2.1claude-2.1Nov 21, 2023 | flagship | Nov 21, 2023 | $8.00 / $24.00 | $8.00 / $24.00 | Jul 21, 2025 |
| Claude 3 Opusclaude-3-opus-20240229Mar 4, 2024 | flagship | Mar 4, 2024 | $15.00 / $75.00 | $15.00 / $75.00 | Jan 5, 2026 |
| Claude 3 Sonnetclaude-3-sonnet-20240229Mar 4, 2024 | mid | Mar 4, 2024 | $3.00 / $15.00 | $3.00 / $15.00 | Jul 21, 2025 |
| Claude 3 Haikuclaude-3-haiku-20240307Mar 13, 2024 | small | Mar 13, 2024 | $0.25 / $1.25 | $0.25 / $1.25 | Apr 20, 2026 |
| Claude 3.5 Sonnetclaude-3-5-sonnet-20240620Jun 20, 2024 | mid | Jun 20, 2024 | $3.00 / $15.00 | $3.00 / $15.00 | Oct 28, 2025 |
| Claude 3.5 Sonnet (upgraded, Oct 2024)claude-3-5-sonnet-20241022Oct 22, 2024 | mid | Oct 22, 2024 | $3.00 / $15.00 | $3.00 / $15.00 | Oct 28, 2025 |
| Claude 3.5 Haikusecondaryclaude-3-5-haiku-20241022Nov 4, 2024 | small | Nov 4, 2024 | $1.00 / $5.00 | $0.80 / $4.00▼ 20% in ▼ 20% out | Feb 19, 2026 |
| Claude 3.7 Sonnetclaude-3-7-sonnet-20250219Feb 24, 2025 | mid | Feb 24, 2025 | $3.00 / $15.00 | $3.00 / $15.00 | Feb 19, 2026 |
| Claude Opus 4claude-opus-4-20250514May 22, 2025 | flagship | May 22, 2025 | $15.00 / $75.00 | $15.00 / $75.00 | Jun 15, 2026 |
| Claude Sonnet 4claude-sonnet-4-20250514May 22, 2025 | mid | May 22, 2025 | $3.00 / $15.00 | $3.00 / $15.00 | Jun 15, 2026 |
| Claude Opus 4.1profile →claude-opus-4-1-20250805Aug 5, 2025 | flagship | Aug 5, 2025 | $15.00 / $75.00 | $15.00 / $75.00 | Aug 5, 2026 |
| Claude Sonnet 4.5not offeredprofile →claude-sonnet-4-5-20250929Sep 29, 2025 | mid | Sep 29, 2025 | $3.00 / $15.00 | $3.00 / $15.00 | still served |
| Claude Haiku 4.5offeredprofile →claude-haiku-4-5-20251001Oct 15, 2025 | small | Oct 15, 2025 | $1.00 / $5.00 | $1.00 / $5.00 | — |
| Claude Opus 4.5not offeredprofile →claude-opus-4-5-20251101Nov 24, 2025 | flagship | Nov 24, 2025 | $5.00 / $25.00 | $5.00 / $25.00 | still served |
| Claude Opus 4.6offeredprofile →claude-opus-4-6Feb 5, 2026 | flagship | Feb 5, 2026 | $5.00 / $25.00 | $5.00 / $25.00 | — |
| Claude Sonnet 4.6offeredprofile →claude-sonnet-4-6Feb 17, 2026 | mid | Feb 17, 2026 | $3.00 / $15.00 | $3.00 / $15.00 | — |
| Claude Opus 4.7offeredprofile →claude-opus-4-7Apr 16, 2026 | flagship | Apr 16, 2026 | $5.00 / $25.00 | $5.00 / $25.00 | — |
| Claude Opus 4.8offeredprofile →claude-opus-4-8May 28, 2026 | flagship | May 28, 2026 | $5.00 / $25.00 | $5.00 / $25.00 | — |
| Claude Fable 5offeredprofile →claude-fable-5Jun 9, 2026 | flagship | Jun 9, 2026 | $10.00 / $50.00 | $10.00 / $50.00 | — |
| Claude Sonnet 5offeredprofile →claude-sonnet-5Jun 30, 2026 | mid | Jun 30, 2026 | $2.00 / $10.00 | $2.00 / $10.00 | — |
| Claude Opus 5offeredprofile →claude-opus-5Jul 24, 2026 | flagship | Jul 24, 2026 | $5.00 / $25.00 | $5.00 / $25.00 | — |
| Claude Fable 5.1offeredprofile →claude-fable-5-1Sep 1, 2026 | flagship | Sep 1, 2026 | $10.00 / $50.00 | $10.00 / $50.00 | — |
| Claude Opus 5.5offeredprofile →claude-opus-5-5Sep 22, 2026 | flagship | Sep 22, 2026 | $4.00 / $20.00 | $4.00 / $20.00 | — |
| Claude Sonnet 5.5offeredprofile →claude-sonnet-5-5Sep 28, 2026 | mid | Sep 28, 2026 | $2.00 / $10.00 | $2.00 / $10.00 | — |
| Claude Haiku 5.5offeredprofile →claude-haiku-5-5Oct 7, 2026 | small | Oct 7, 2026 | $0.10 / $0.50 | $0.10 / $0.50 | — |
Gemini · Google API
Verified Nov 2, 2026 against the Gemini API pricing pageSince July: 4 new models (Gemini 4 Argon, Gemini 3.8 Flash, Gemini 3.6 Flash…) · 1 price cut (Gemini 3.6 Flash) · 3 announced future changes.
| Model | Input / 1M | Output / 1M | Blended 3:1 | Since launch | Last change |
|---|---|---|---|---|---|
| Gemini 4 Argon synthesis intro → Nov 30, 2026 profile → gemini-4-argon ▲ $4.00 / $20.00 from Dec 1, 2026 | $2.00 | $10.00 | $4.00 | unchangedsince Oct 2026 | new · Oct 13, 2026 |
| Gemini 3.1 Pro (preview) profile → gemini-3.1-pro-preview | $2.00 | $12.00 | $4.50 | unchangedsince Feb 2026 | — |
| Gemini 3.8 Flash default intro → Dec 31, 2026 profile → gemini-3.8-flash ▲ $1.50 / $7.50 from Jan 1, 2027 | $0.75 | $3.75 | $1.50 | unchangedsince Sep 2026 | — |
| Gemini 3.6 Flash promo → Dec 31, 2026 profile → gemini-3.6-flash ▲ $1.50 / $7.50 from Jan 1, 2027 | $0.75 | $3.75 | $1.50 | ▼ 50% in▼ 50% outsince Jul 2026 | ▼ Aug 13, 2026 |
| Gemini 2.5 Flash-Lite sample unlisted by Google profile → gemini-2.5-flash-lite | $0.10 | $0.40 | $0.175 | unchangedsince Jul 2025 | — |
| Gemini 3 Flash (preview) more profile → gemini-3-flash-preview | $0.50 | $3.00 | $1.125 | unchangedsince Dec 2025 | — |
| Gemini 3.5 Flash-Lite more profile → gemini-3.5-flash-lite | $0.30 | $2.50 | $0.85 | unchangedsince Jul 2026 | — |
| Gemini 3.1 Flash-Lite more profile → gemini-3.1-flash-lite | $0.25 | $1.50 | $0.563 | unchangedsince May 2026 | — |
| No models match the current filters. | |||||
Gemini — price history
Jan 1, 2027
Gemini 3.8 FlashScheduled increaseannouncedin$0.75$1.50▲ 100%out$3.75$7.50▲ 100%
Google: “$1.50 / $7.50 starting January 1, 2027.” source
Dec 1, 2026
Gemini 4 ArgonScheduled increaseannouncedin$2.00$4.00▲ 100%out$10.00$20.00▲ 100%
Google: “$4/1M input and $20/1M output thereafter” (from December 1, 2026). source
Oct 13, 2026
Gemini 4 ArgonLaunched at an introductory priceuntil Nov 30, 2026in$2.00out$10.00
Launched GA at Google’s introductory rate, “through Nov 30, 2026”. No prompt-length tier. source
Sep 5, 2026
Gemini 2.5 Flash-LitePrice no longer listedlast listed$0.10 / $0.40
Google removed the model from its pricing page and, from Sep 6, closed it to new Google Cloud projects; existing keys are still served at the last listed price. source
Sep 2, 2026
Gemini 3.8 FlashLaunched at an introductory priceuntil Dec 31, 2026in$0.75out$3.75
Launched at Google’s introductory Flash rate, “through December 31, 2026.” source
Sep 2, 2026
Gemini 3.6 FlashPolyCog’s estimate
PolyCog’s estimate stayed at $1.50 / $7.50 from Aug 13 until Sep 2, 2026 (over-estimated 2×).
Aug 13, 2026
Gemini 3.6 FlashPromotional priceuntil Dec 31, 2026secondary sourcein$1.50$0.75▼ 50%out$7.50$3.75▼ 50%
Moved onto the introductory Flash rate the day 3.7 Flash launched. No changelog entry; dated from pricing-page captures. source
Aug 13, 2026
Gemini 3.7 FlashLaunched at an introductory pricenot offereduntil Dec 31, 2026in$0.75out$3.75
Launched “at an introductory price through December 31, 2026.” source
May 19, 2026
Gemini 3.5 FlashLaunchednot offeredin$1.50out$9.00
Google now labels it a legacy model; not covered by the 3.6–3.8 introductory rate. source
May 7, 2026
Gemini 3.1 Flash-LiteLaunchedin$0.25out$1.50
Preview Mar 3, 2026. Google’s earliest shutdown date: May 7, 2027. source
Feb 19, 2026
Gemini 3.1 Pro (preview)Launchedin$2.00out$12.00
Prompts over 200K tokens bill $4 / $18. source
Dec 17, 2025
Gemini 3 Flash (preview)Launchedin$0.50out$3.00
Google now calls it its legacy Flash model. source
Nov 18, 2025
Gemini 3 Pro PreviewLaunchedretired Mar 2026secondary sourcein$2.00out$12.00
Originally priced as $2.00 / $12.00 per 1,000,000 tokens for prompts <=200K tokens ($4.00 / $18.00 above 200K). source
Jun 17, 2025
Gemini 2.5 ProIntroductory price made permanentnot offeredin$1.25out$10.00
Reached general availability (plain 'gemini-2.5-pro' id) at unchanged pricing from the paid preview. source
≈Jun 17, 2025
Gemini 2.5 FlashPrice increasenot offeredin$0.15$0.30▲ 100%out$0.60$2.50▲ 317%
At GA, pricing was revised upward to a single unified rate (both thinking and non-thinking output billed the same), replacing the preview's cheap non-thinking-output / $3.50-thinking split. Net effect: 2x higher input, higher effective output for non-thinking source
Apr 17, 2025
Gemini 2.5 FlashLaunchednot offeredsecondary sourcein$0.15out$0.60
Originally priced as $0.15 / $0.60 per 1,000,000 tokens (non-thinking output; output priced at $3.50/1M when the thinking budget is used). source
Apr 4, 2025
Gemini 2.5 ProLaunchednot offeredsecondary sourcein$1.25out$10.00
Originally priced as $1.25 / $10.00 per 1,000,000 tokens for prompts <=200K tokens ($2.50 / $15.00 above 200K). source
Feb 25, 2025
Gemini 2.0 Flash-LiteLaunchedretired Jun 2026secondary sourcein$0.075out$0.30
Originally priced as $0.075 / $0.30 per 1,000,000 tokens (single tier). source
Feb 5, 2025
Gemini 2.0 FlashLaunchedretired Jun 2026in$0.10out$0.40
Originally priced as $0.10 / $0.40 per 1,000,000 tokens (single tier, no long-context split). source
Oct 3, 2024
Gemini 1.5 Flash-8BLaunchedretired Sep 2025in$0.038out$0.15
Originally priced as $0.0375 / $0.15 per 1,000,000 tokens for prompts <=128K tokens. source
Oct 1, 2024
Gemini 1.5 ProPrice cutretired Sep 2025in$3.50$1.25▼ 64%out$10.50$5.00▼ 52%
64% cut on input tokens, 52% cut on output tokens (Google's own percentages) for prompts <=128K, announced 2024-09-24 and effective 2024-10-01, alongside the gemini-1.5-pro-002 model refresh and higher (1,000 RPM) paid-tier rate limits. source
Aug 12, 2024
Gemini 1.5 FlashPrice cutretired Sep 2025in$0.35$0.075▼ 79%out$1.05$0.30▼ 71%
78% cut on input, 71% cut on output, for prompts <=128K tokens, with comparable cuts to the >128K tier and to context caching. source
May 23, 2024
Gemini 1.5 ProLaunchedretired Sep 2025secondary sourcein$3.50out$10.50
Originally priced as $3.50 / $10.50 per 1,000,000 tokens for prompts <=128K tokens. source
May 23, 2024
Gemini 1.5 FlashLaunchedretired Sep 2025secondary sourcein$0.35out$1.05
Originally priced as $0.35 / $1.05 per 1,000,000 tokens for prompts <=128K tokens. source
≈Feb 15, 2024
Gemini 1.0 Pro (Gemini Pro)Launchedretired Feb 2025in$0.50out$1.50
Originally priced as $0.000125 per 1,000 characters (input) / $0.000375 per 1,000 characters (output) — converted to per-1,000,000-tokens at . source
No dated history for this selection yet.
Since launch
Google API since Dec 2023 · first public price Feb 2024Flagship: Gemini 1.0 Pro (Gemini Pro) at $0.50 / $1.50 → Gemini 4 Argon at $2.00 / $10.00 today. Cheapest: Gemini 1.0 Pro (Gemini Pro) at $0.50 / $1.50 → Gemini 2.5 Flash-Lite at $0.10 / $0.40.
Flagship — the provider’s best model at the timeCheapest text modelblended 3 in : 1 out · USD per 1M tokens · log scale
Hover or tap for the exact input / output at any change · ← → on a focused chart · 12 dated changes
All 19 Google models since Feb 2024 — launch and last price
| Model | Tier | Launched | Launch in / out | Last in / out | Retired |
|---|---|---|---|---|---|
| Gemini 1.0 Pro (Gemini Pro)≈ dategemini-1.0-pro≈ Feb 15, 2024 | flagship | ≈ Feb 15, 2024 | $0.50 / $1.50 | $0.50 / $1.50 | Feb 18, 2025 |
| Gemini 1.5 Prosecondarygemini-1.5-proMay 23, 2024 | flagship | May 23, 2024 | $3.50 / $10.50 | $1.25 / $5.00▼ 64% in ▼ 52% out | Sep 29, 2025 |
| Gemini 1.5 Flashsecondarygemini-1.5-flashMay 23, 2024 | mid | May 23, 2024 | $0.35 / $1.05 | $0.075 / $0.30▼ 79% in ▼ 71% out | Sep 29, 2025 |
| Gemini 1.5 Flash-8Bgemini-1.5-flash-8bOct 3, 2024 | small | Oct 3, 2024 | $0.038 / $0.15 | $0.038 / $0.15 | Sep 29, 2025 |
| Gemini 2.0 Flashgemini-2.0-flashFeb 5, 2025 | mid | Feb 5, 2025 | $0.10 / $0.40 | $0.10 / $0.40 | Jun 1, 2026 |
| Gemini 2.0 Flash-Litesecondarygemini-2.0-flash-liteFeb 25, 2025 | small | Feb 25, 2025 | $0.075 / $0.30 | $0.075 / $0.30 | Jun 1, 2026 |
| Gemini 2.5 Pronot offeredsecondaryprofile →gemini-2.5-proApr 4, 2025 | flagship | Apr 4, 2025 | $1.25 / $10.00 | $1.25 / $10.00 | still served |
| Gemini 2.5 Flashnot offeredsecondaryprofile →gemini-2.5-flashApr 17, 2025 | mid | Apr 17, 2025 | $0.15 / $0.60 | $0.30 / $2.50▲ 100% in ▲ 317% out | still served |
| Gemini 2.5 Flash-Liteofferedprofile →gemini-2.5-flash-liteJul 22, 2025 | small | Jul 22, 2025 | $0.10 / $0.40 | $0.10 / $0.40 | — |
| Gemini 3 Pro Previewsecondarygemini-3-pro-previewNov 18, 2025 | flagship | Nov 18, 2025 | $2.00 / $12.00 | $2.00 / $12.00 | Mar 9, 2026 |
| Gemini 3 Flash (preview)offeredprofile →gemini-3-flash-previewDec 17, 2025 | mid | Dec 17, 2025 | $0.50 / $3.00 | $0.50 / $3.00 | — |
| Gemini 3.1 Pro (preview)offeredprofile →gemini-3.1-pro-previewFeb 19, 2026 | flagship | Feb 19, 2026 | $2.00 / $12.00 | $2.00 / $12.00 | — |
| Gemini 3.1 Flash-Liteofferedprofile →gemini-3.1-flash-liteMay 7, 2026 | small | May 7, 2026 | $0.25 / $1.50 | $0.25 / $1.50 | — |
| Gemini 3.5 Flashnot offeredprofile →gemini-3.5-flashMay 19, 2026 | mid | May 19, 2026 | $1.50 / $9.00 | $1.50 / $9.00 | still served |
| Gemini 3.6 Flashofferedprofile →gemini-3.6-flashJul 21, 2026 | mid | Jul 21, 2026 | $1.50 / $7.50 | $0.75 / $3.75▼ 50% in ▼ 50% out | — |
| Gemini 3.5 Flash-Liteofferedprofile →gemini-3.5-flash-liteJul 21, 2026 | small | Jul 21, 2026 | $0.30 / $2.50 | $0.30 / $2.50 | — |
| Gemini 3.7 Flashnot offeredprofile →gemini-3.7-flashAug 13, 2026 | mid | Aug 13, 2026 | $0.75 / $3.75 | $0.75 / $3.75 | still served |
| Gemini 3.8 Flashofferedprofile →gemini-3.8-flashSep 2, 2026 | mid | Sep 2, 2026 | $0.75 / $3.75 | $0.75 / $3.75 | — |
| Gemini 4 Argonofferedprofile →gemini-4-argonOct 13, 2026 | flagship | Oct 13, 2026 | $2.00 / $10.00 | $2.00 / $10.00 | — |
DeepSeek · DeepSeek API
Verified Nov 2, 2026 against the DeepSeek pricing pageSince July: 1 new model (DeepSeek V4.1 Flash) · 1 price cut (DeepSeek V4 Pro) · 1 increase (DeepSeek V4 Pro).
| Model | Input / 1M | Output / 1M | Blended 3:1 | Since launch | Last change |
|---|---|---|---|---|---|
| DeepSeek V4 Pro synthesis profile → deepseek-v4-pro | $1.32off-peak $0.66 | $3.96off-peak $1.98 | $1.98 | ▼ 24% in▲ 14% outsince ≈ Apr 2026 | ▲ Aug 16, 2026 |
| DeepSeek V4.1 Flash default sample profile → deepseek-flash | $0.30off-peak $0.15 | $1.20off-peak $0.60 | $0.525 | unchangedsince Sep 2026 | — |
| No models match the current filters. | |||||
DeepSeek — price history
Sep 11, 2026
DeepSeek V4.1 FlashPolyCog’s estimate
Estimated at the peak rate around the clock (never under-estimates); off-peak actuals settle in your DeepSeek bill.
Sep 10, 2026
DeepSeek V4.1 FlashLaunchedin$0.30out$1.20off-peak$0.15 / $0.60
DeepSeek-V4.1-Flash under the new id deepseek-flash ("Change the model name to `deepseek-flash` to call the latest V4.1 Flash model"). Cache-hit input $0.006 peak / $0.003 off-peak; 1M context, 384K max output, native vision, thinking and non-thinking modes. New pricing effective 04:00 UTC, Sep 10, 2026; same peak / off-peak windows as V4 (Mon-Fri 01:00-04:00 and 06:00-10:00 UTC peak, 50% off otherwise). source
Sep 10, 2026
DeepSeek V4 FlashPrice cutretired Sep 2026in$0.44$0.30▼ 32%out$1.32$1.20▼ 9%off-peak$0.15 / $0.60
DeepSeek-V4-Flash retired with the V4.1 Flash launch; the deepseek-v4-flash id is "temporarily routed to V4.1 Flash" and billed at the Flash price. PolyCog moved its catalog and sample tier to the new deepseek-flash id the same week; this row stays for sessions saved before the switch. source
Aug 18, 2026
DeepSeek V4 ProPolyCog’s estimate
PolyCog estimates at the peak rate around the clock (it never under-estimates); off-peak actuals settle in your provider bill.
Aug 16, 2026
DeepSeek V4 ProPrice increasein$0.435$1.32▲ 203%out$0.87$3.96▲ 355%off-peak$0.66 / $1.98
New peak / off-peak split from 16:00 UTC, Aug 16, 2026. Peak = Mon–Fri 09:00–12:00 and 14:00–18:00 Beijing time (01:00–04:00 and 06:00–10:00 UTC); off-peak is 50% lower. V4 Pro reached GA Aug 13. source
Aug 16, 2026
DeepSeek V4 FlashPrice increaseretired Sep 2026in$0.14$0.44▲ 214%out$0.28$1.32▲ 371%off-peak$0.22 / $0.66
Same peak / off-peak split as V4 Pro. source
≈Jul 2026
DeepSeek V4 ProPrice cutin$1.74$0.435▼ 75%out$3.48$0.87▼ 75%
About 75% off. Exact date not published — recorded by Jul 25, 2026.
≈Apr 24, 2026
DeepSeek V4 ProLaunchedin$1.74out$3.48
V4 preview API opened Apr 24, 2026; price as recorded in June 2026 (DeepSeek publishes its table as an image). source
≈Apr 24, 2026
DeepSeek V4 FlashLaunchedretired Sep 2026in$0.14out$0.28
Input price as recorded by PolyCog; output confirmed by press coverage of the August increase. source
Apr 24, 2026
DeepSeek Chat (V2 → V2.5 → V3 → V3.1 → V3.2)Price cutretired Jul 2026in$0.28$0.14▼ 50%out$0.42$0.28▼ 33%
DeepSeek-V4 launched; the legacy 'deepseek-chat' id was temporarily remapped to DeepSeek-V4-Flash's non-thinking mode at V4-Flash's launch price, pending full retirement. source
Apr 24, 2026
DeepSeek Reasoner (R1 → R1-0528 → V3.1/V3.2 thinking mode)Price cutretired Jul 2026in$0.28$0.14▼ 50%out$0.42$0.28▼ 33%
DeepSeek-V4 launched; the legacy 'deepseek-reasoner' id was temporarily remapped to DeepSeek-V4-Flash's thinking mode at V4-Flash's launch price, pending full retirement. source
Sep 29, 2025
DeepSeek Chat (V2 → V2.5 → V3 → V3.1 → V3.2)Price cutretired Jul 2026in$0.56$0.28▼ 50%out$1.68$0.42▼ 75%
DeepSeek-V3.2-Exp launched with DeepSeek Sparse Attention; cache-hit input fell to $0.028/1M. Described by DeepSeek as a 50%+ reduction from V3.1/V3.1-Terminus pricing. source
Sep 29, 2025
DeepSeek Reasoner (R1 → R1-0528 → V3.1/V3.2 thinking mode)Price cutretired Jul 2026in$0.56$0.28▼ 50%out$1.68$0.42▼ 75%
DeepSeek-V3.2-Exp launched; deepseek-reasoner (thinking mode) priced identically to deepseek-chat. source
Sep 5, 2025
DeepSeek Chat (V2 → V2.5 → V3 → V3.1 → V3.2)Price increaseretired Jul 2026in$0.27$0.56▲ 107%out$1.10$1.68▲ 53%
DeepSeek-V3.1 (hybrid thinking/non-thinking architecture) launched 2025-08-21; unified pricing for deepseek-chat and deepseek-reasoner took effect and the Feb 2025 off-peak discount program ended, both at 16:00 UTC on this date. Cache-hit input stayed $0.07/1M source
Sep 5, 2025
DeepSeek Reasoner (R1 → R1-0528 → V3.1/V3.2 thinking mode)Price cutretired Jul 2026in$0.55$0.56▲ 2%out$2.19$1.68▼ 23%
Unified with deepseek-chat under the V3.1 hybrid architecture (launched 2025-08-21) and the off-peak discount program ended, both at 16:00 UTC. Cache-hit input halved ($0.14→$0.07); cache-miss input rose $0.01 (effectively flat, $0.55→$0.56); output fell ~23% source
Feb 26, 2025
DeepSeek Chat (V2 → V2.5 → V3 → V3.1 → V3.2)Promotional priceretired Jul 2026until Sep 5, 2025in$0.27out$1.10
Off-peak discount program begins: DeepSeek-V3 (deepseek-chat) at 50% off, daily 16:30-00:30 UTC. Standard/peak rate unchanged. Program ran until 2025-09-05. source
Feb 26, 2025
DeepSeek Reasoner (R1 → R1-0528 → V3.1/V3.2 thinking mode)Promotional priceretired Jul 2026until Sep 5, 2025in$0.55out$2.19
Off-peak discount program begins: DeepSeek-R1 at 75% off, daily 16:30-00:30 UTC. Standard/peak rate unchanged. Program ran until 2025-09-05. source
Feb 9, 2025
DeepSeek Chat (V2 → V2.5 → V3 → V3.1 → V3.2)Price increaseretired Jul 2026secondary sourcein$0.14$0.27▲ 93%out$0.28$1.10▲ 293%
Promotional V3 pricing ended (45-day promo begun 2024-12-26); cache-hit input rose to $0.07/1M. Press reported the output figure as $1.09-$1.10/1M depending on rounding of the underlying ¥8 rate. source
Jan 20, 2025
DeepSeek Reasoner (R1 → R1-0528 → V3.1/V3.2 thinking mode)Launchedretired Jul 2026in$0.55out$2.19
Dec 26, 2024
DeepSeek Chat (V2 → V2.5 → V3 → V3.1 → V3.2)Promotional priceretired Jul 2026until Feb 8, 2025in$0.14out$0.28
deepseek-chat upgraded in place to DeepSeek-V3; DeepSeek explicitly kept the same rate as V2 through a promotional window running to 2025-02-08. source
May 6, 2024
DeepSeek Chat (V2 → V2.5 → V3 → V3.1 → V3.2)Launchedretired Jul 2026secondary sourcein$0.14out$0.28
Originally priced as ¥1 per 1M input tokens / ¥2 per 1M output tokens (no caching tier existed yet). source
No dated history for this selection yet.
Since launch
DeepSeek API since May 2024 · first public price May 2024Flagship: DeepSeek Chat (V2 → V2.5 → V3 → V3.1 → V3.2) at $0.14 / $0.28 → DeepSeek V4 Pro at $1.32 / $3.96 today. Cheapest: DeepSeek Chat (V2 → V2.5 → V3 → V3.1 → V3.2) at $0.14 / $0.28 → DeepSeek V4.1 Flash at $0.30 / $1.20.
Flagship — the provider’s best model at the timeCheapest text modelblended 3 in : 1 out · USD per 1M tokens · log scale
Hover or tap for the exact input / output at any change · ← → on a focused chart · 11 dated changes
All 5 DeepSeek models since May 2024 — launch and last price
| Model | Tier | Launched | Launch in / out | Last in / out | Retired |
|---|---|---|---|---|---|
| DeepSeek Chat (V2 → V2.5 → V3 → V3.1 → V3.2)secondaryprofile →deepseek-chatMay 6, 2024 | flagship | May 6, 2024 | $0.14 / $0.28 | $0.14 / $0.28 | Jul 24, 2026 |
| DeepSeek Reasoner (R1 → R1-0528 → V3.1/V3.2 thinking mode)profile →deepseek-reasonerJan 20, 2025 | reasoning | Jan 20, 2025 | $0.55 / $2.19 | $0.14 / $0.28▼ 75% in ▼ 87% out | Jul 24, 2026 |
| DeepSeek V4 Prooffered≈ dateprofile →deepseek-v4-pro≈ Apr 24, 2026 | flagship | ≈ Apr 24, 2026 | $1.74 / $3.48 | $1.32 / $3.96▼ 24% in ▲ 14% out | — |
| DeepSeek V4 Flash≈ dateprofile →deepseek-v4-flash≈ Apr 24, 2026 | small | ≈ Apr 24, 2026 | $0.14 / $0.28 | $0.30 / $1.20▲ 114% in ▲ 329% out | Sep 10, 2026 |
| DeepSeek V4.1 Flashofferedprofile →deepseek-flashSep 10, 2026 | small | Sep 10, 2026 | $0.30 / $1.20 | $0.30 / $1.20 | — |
Grok · xAI API
Verified Nov 2, 2026 against the xAI pricing pageSince July: 3 new models (Grok 4.7, Grok 4.6, Grok 4.5).
| Model | Input / 1M | Output / 1M | Blended 3:1 | Since launch | Last change |
|---|---|---|---|---|---|
| Grok 4.7 flagship profile → grok-4.7 | $2.00 | $6.00 | $3.00 | unchangedsince Sep 2026 | — |
| Grok 4.6 profile → grok-4.6 | $2.00 | $6.00 | $3.00 | unchangedsince Aug 2026 | — |
| Grok 4.5 profile → grok-4.5 | $2.00 | $6.00 | $3.00 | unchangedsince Jul 2026 | — |
| Grok 4.3 default profile → grok-4.3 | $1.25 | $2.50 | $1.563 | unchangedsince May 2026 | — |
| No models match the current filters. | |||||
Grok — price history
Sep 21, 2026
Grok 4.7Launchedin$2.00out$6.00
Cached input $0.50. Prompts of 200K+ tokens bill $4 / $12 for the whole request. Grok 4.7 Fast (2× the token rates) is available only through Cursor and Grok Build, not the API. source
Aug 12, 2026
Grok 4.6Launchedin$2.00out$6.00
Prompts of 200K+ tokens bill $4 / $12 for the whole request. source
May 1, 2026
Grok 4.3Launchedsecondary sourcein$1.25out$2.50
Dated from launch coverage (Apr 30 / May 1, 2026); not on x.ai’s news index.
Nov 4, 2024
Grok BetaPromotional pricenot offeredin$5.00out$15.00
xAI's official 'API Public Beta' post reconfirmed grok-beta's price/128K context and introduced $25/month in free API credits for every account (prepaid holders got matching credits, e.g. a $50 top-up yielded $75/month), running through the end of December 202 source
No dated history for this selection yet.
Since launch
xAI API since Oct 2024 · first public price Oct 2024Flagship: Grok Beta at $5.00 / $15.00 → Grok 4.7 at $2.00 / $6.00 today. Cheapest: Grok Beta at $5.00 / $15.00 → Grok 4.3 at $1.25 / $2.50.
Flagship — the provider’s best model at the timeCheapest text modelblended 3 in : 1 out · USD per 1M tokens · log scale
Hover or tap for the exact input / output at any change · ← → on a focused chart · 13 dated changes
All 20 xAI models since Oct 2024 — launch and last price
| Model | Tier | Launched | Launch in / out | Last in / out | Retired |
|---|---|---|---|---|---|
| Grok Betanot offeredsecondarygrok-betaOct 21, 2024 | flagship | Oct 21, 2024 | $5.00 / $15.00 | $5.00 / $15.00 | still served |
| Grok 2 (1212)not offeredgrok-2-1212Dec 14, 2024 | flagship | Dec 14, 2024 | $2.00 / $10.00 | $2.00 / $10.00 | still served |
| Grok 3secondarygrok-3Apr 9, 2025 | flagship | Apr 9, 2025 | $3.00 / $15.00 | $3.00 / $15.00 | May 15, 2026 |
| Grok 3 Fastnot offeredsecondarygrok-3-fastApr 9, 2025 | mid | Apr 9, 2025 | $5.00 / $25.00 | $5.00 / $25.00 | still served |
| Grok 3 Mininot offeredsecondarygrok-3-miniApr 9, 2025 | reasoning | Apr 9, 2025 | $0.30 / $0.50 | $0.30 / $0.50 | still served |
| Grok 3 Mini Fastnot offeredsecondarygrok-3-mini-fastApr 9, 2025 | reasoning | Apr 9, 2025 | $0.60 / $4.00 | $0.60 / $4.00 | still served |
| Grok 4secondarygrok-4Jul 9, 2025 | flagship | Jul 9, 2025 | $3.00 / $15.00 | $3.00 / $15.00 | May 15, 2026 |
| Grok Code Fast 1grok-code-fast-1Aug 28, 2025 | other | Aug 28, 2025 | $0.20 / $1.50 | $0.20 / $1.50 | May 15, 2026 |
| Grok 4 Fast (Reasoning)grok-4-fast-reasoningSep 19, 2025 | reasoning | Sep 19, 2025 | $0.20 / $0.50 | $0.20 / $0.50 | May 15, 2026 |
| Grok 4 Fast (Non-Reasoning)grok-4-fast-non-reasoningSep 19, 2025 | mid | Sep 19, 2025 | $0.20 / $0.50 | $0.20 / $0.50 | May 15, 2026 |
| Grok 4.1secondarygrok-4-1Nov 17, 2025 | flagship | Nov 17, 2025 | $3.00 / $15.00 | $3.00 / $15.00 | May 15, 2026 |
| Grok 4.1 Fast (Reasoning)secondarygrok-4-1-fast-reasoningNov 19, 2025 | reasoning | Nov 19, 2025 | $0.20 / $0.50 | $0.20 / $0.50 | May 15, 2026 |
| Grok 4.1 Fast (Non-Reasoning)grok-4-1-fast-non-reasoningNov 19, 2025 | mid | Nov 19, 2025 | $0.20 / $0.50 | $0.20 / $0.50 | May 15, 2026 |
| Grok 4.20 (Reasoning)not offeredgrok-4.20-0309-reasoningMar 10, 2026 | reasoning | Mar 10, 2026 | $1.25 / $2.50 | $1.25 / $2.50 | still served |
| Grok 4.20 (Non-Reasoning)not offeredgrok-4.20-0309-non-reasoningMar 10, 2026 | mid | Mar 10, 2026 | $1.25 / $2.50 | $1.25 / $2.50 | still served |
| Grok 4.3offeredsecondaryprofile →grok-4.3May 1, 2026 | flagship | May 1, 2026 | $1.25 / $2.50 | $1.25 / $2.50 | — |
| Grok Build 0.1not offeredgrok-build-0.1May 19, 2026 | other | May 19, 2026 | $1.00 / $2.00 | $1.00 / $2.00 | still served |
| Grok 4.5offeredprofile →grok-4.5Jul 16, 2026 | flagship | Jul 16, 2026 | $2.00 / $6.00 | $2.00 / $6.00 | — |
| Grok 4.6offeredprofile →grok-4.6Aug 12, 2026 | flagship | Aug 12, 2026 | $2.00 / $6.00 | $2.00 / $6.00 | — |
| Grok 4.7offeredprofile →grok-4.7Sep 21, 2026 | flagship | Sep 21, 2026 | $2.00 / $6.00 | $2.00 / $6.00 | — |
Wildcard · OpenRouter API
Verified Nov 2, 2026 against the OpenRouter model listingSince July: 10 new models (Qwen3.8 Max, Qwen3.8 Flash, Kimi K3.1 Pro…) · 1 price cut (Mistral Large 4).
| Model | Input / 1M | Output / 1M | Blended 3:1 | Since launch | Last change |
|---|---|---|---|---|---|
| Qwen3.8 Max flagship profile → qwen/qwen3.8-max-0902 | $2.00 | $6.00 | $3.00 | unchangedsince Sep 2026 | — |
| Qwen3.8 Flash default profile → qwen/qwen3.8-flash | $0.15 | $0.47 | $0.23 | unchangedsince Aug 2026 | — |
| Kimi K3.1 Pro profile → moonshotai/kimi-k3.1-pro | $2.00 | $10.00 | $4.00 | unchangedsince Oct 2026 | new · Oct 23, 2026 |
| Kimi K3 profile → moonshotai/kimi-k3 | $3.00 | $15.00 | $6.00 | unchangedsince Jul 2026 | — |
| Kimi K2.6 profile → moonshotai/kimi-k2.6 | $0.95 | $4.00 | $1.713 | unchangedsince Apr 2026 | — |
| GLM 5.3 profile → z-ai/glm-5.3 | $1.40 | $4.40 | $2.15 | unchangedsince Aug 2026 | — |
| GLM 5.3 Flash profile → z-ai/glm-5.3-flash | $0.15 | $0.50 | $0.238 | unchangedsince Aug 2026 | — |
| Mistral Large 4 profile → mistralai/mistral-large-4-0 | $0.70 | $2.10 | $1.05 | ▼ 49% in▼ 50% outsince Oct 2026 | ▼ ≈ Nov 2, 2026 |
| Mistral Medium 3.5 profile → mistralai/mistral-medium-3-5 | $1.50 | $7.50 | $3.00 | unchangedsince Apr 2026 | — |
| Mistral Small 4 profile → mistralai/mistral-small-2603 | $0.15 | $0.60 | $0.263 | unchangedsince Mar 2026 | — |
| Llama 4 Maverick profile → meta-llama/llama-4-maverick | $0.188 | $0.653 | $0.304 | unchangedsince Apr 2025 | — |
| Muse Wonder profile → meta/muse-wonder | $4.00 | $20.00 | $8.00 | unchangedsince Oct 2026 | new · Oct 12, 2026 |
| Muse Spark 1.3 profile → meta/muse-spark-1.3 | $1.25 | $4.25 | $2.00 | unchangedsince Sep 2026 | — |
| MiniMax M3 profile → minimax/minimax-m3 | $0.30 | $1.20 | $0.525 | unchangedsince May 2026 | — |
| MiniMax M3.1 Flash profile → minimax/minimax-m3.1-flash | $0.15 | $0.60 | $0.263 | unchangedsince Oct 2026 | new · Oct 19, 2026 |
| Venice Uncensored profile → cognitivecomputations/dolphin-mistral-24b-venice-edition | $0.20 | $0.90 | $0.375 | unchangedsince Jul 2025 | — |
| Hermes 4 405B profile → nousresearch/hermes-4-405b | $1.00 | $3.00 | $1.50 | unchangedsince Aug 2025 | — |
| No models match the current filters. | |||||
Wildcard — price history
≈Nov 2, 2026
Mistral Large 4Price cutin$1.36$0.70▼ 49%out$4.18$2.10▼ 50%
Mistral's model page and its own OpenRouter endpoints now list $0.70 / $2.10 (cached $0.07), with no sale wording and no discount on the endpoint (read 2026-11-02). The list price was $1.36 / $4.18 when last read on 2026-10-26, so the cut landed between those two reads; dated the day it was read. source
Oct 23, 2026
Kimi K3.1 ProLaunchedin$2.00out$10.00
Moonshot's post-trained Kimi K3 (1M context; open weights promised by Nov 4). Moonshot AI's own endpoint (moonshotai/mxfp4) is the only host on OpenRouter, $2 / $10, discount 0; its "highspeed" priority endpoint lists $4 / $20. Read 2026-10-26. source
Oct 19, 2026
MiniMax M3.1 FlashLaunchedin$0.15out$0.60
MiniMax's faster M3 (same ~428B / 23B MoE, sparse attention, quantization-aware training; 1M context). The Minimax endpoint lists $0.15 / $0.60, as do most hosts (SambaNova $0.30 / $1.20). Read 2026-10-26. source
Oct 12, 2026
Muse WonderLaunchedin$4.00out$20.00
Meta Superintelligence Labs' frontier reasoning model (closed weights; Meta's API id muse-wonder-1.0). Meta's own endpoint is the only host on OpenRouter; Meta's pricing page: $4 / $20, cached input $0.20, "No long-context premium". source
Oct 6, 2026
Mistral Large 4Launchedin$1.36out$4.18
Mistral's list price (public preview). Launched on sale at $0.68 / $2.09 with no end date; PolyCog estimates with the list price (the OpenRouter rate rule: the lab's own endpoint, never the day's discount). source
Sep 3, 2026
Qwen3.8 MaxLaunchedsecondary sourcein$2.00out$6.00
OpenRouter listing (1M context); rate read on the model page 2026-09-17; the rate PolyCog estimates with — OpenRouter routes cheapest-host-first with fallbacks up the price ladder (proxy.php, PC84), so a fallback host can bill above it for a turn. source
Sep 2, 2026
Muse Spark 1.3Launchedin$1.25out$4.25
Meta's mid-tier Muse reasoning model (closed weights). Meta's own endpoint is the only host; OpenRouter listed it 2026-09-02 (the endpoint's created date). Rate read on Meta's pricing page and the OpenRouter endpoint 2026-10-13; the -contributor variant (Meta may train on its prompts) is not offered. source
Aug 26, 2026
Qwen3.8 FlashLaunchedsecondary sourcein$0.15out$0.47
OpenRouter listing (1M context); rate read on the model page 2026-09-17; the rate PolyCog estimates with — OpenRouter routes cheapest-host-first with fallbacks up the price ladder (proxy.php, PC84), so a fallback host can bill above it for a turn. source
Aug 26, 2026
GLM 5.3 FlashLaunchedsecondary sourcein$0.15out$0.50
Launch price restated at Z.ai's own endpoint rate on OpenRouter ($0.15 / $0.50, read 2026-09-23; the PC88 rate rule). When PolyCog first read the listing on 2026-09-17 it carried a cheaper host's $0.075 / $0.25; the move to the lab's rate was PolyCog's change of rate basis, not a price change by the lab, so no increase is recorded (PC199, 2026-11-02). source
Aug 18, 2026
GLM 5.3Launchedsecondary sourcein$1.40out$4.40
Launch price restated at Z.ai's own endpoint rate on OpenRouter ($1.40 / $4.40, read 2026-09-23; the PC88 rate rule). When PolyCog first read the listing on 2026-09-17 it carried a cheaper host's $0.90 / $3.00; the move to the lab's rate was PolyCog's change of rate basis, not a price change by the lab, so no increase is recorded (PC199, 2026-11-02). source
Jul 16, 2026
Kimi K3Launchedsecondary sourcein$3.00out$15.00
Launch price restated at Moonshot AI's own endpoint rate on OpenRouter ($3.00 / $15.00, read 2026-09-23; the PC88 rate rule). When PolyCog first read the listing on 2026-09-17 it carried a cheaper host's $2.10 / $10.95; the move to the lab's rate was PolyCog's change of rate basis, not a price change by the lab, so no increase is recorded (PC199, 2026-11-02). source
May 31, 2026
MiniMax M3Launchedsecondary sourcein$0.30out$1.20
Launch price restated at MiniMax's own endpoint rate on OpenRouter ($0.30 / $1.20, read 2026-09-29; the PC88 rate rule). When PolyCog first read the listing on 2026-09-17 it carried a cheaper host's $0.23 / $0.96; the move to the lab's rate was PolyCog's change of rate basis, not a price change by the lab, so no increase is recorded (PC199, 2026-11-02). source
Apr 30, 2026
Mistral Medium 3.5Launchedsecondary sourcein$1.50out$7.50
OpenRouter listing (262K context); rate read on the model page 2026-09-17; the rate PolyCog estimates with — OpenRouter routes cheapest-host-first with fallbacks up the price ladder (proxy.php, PC84), so a fallback host can bill above it for a turn. source
Apr 20, 2026
Kimi K2.6Launchedsecondary sourcein$0.95out$4.00
Launch price restated at Moonshot AI's own endpoint rate on OpenRouter ($0.95 / $4.00, read 2026-09-23; the PC88 rate rule). When PolyCog first read the listing on 2026-09-17 it carried a cheaper host's $0.471 / $2.835; the move to the lab's rate was PolyCog's change of rate basis, not a price change by the lab, so no increase is recorded (PC199, 2026-11-02). source
Mar 16, 2026
Mistral Small 4Launchedsecondary sourcein$0.15out$0.60
OpenRouter listing (262K context); rate read on the model page 2026-09-17; the rate PolyCog estimates with — OpenRouter routes cheapest-host-first with fallbacks up the price ladder (proxy.php, PC84), so a fallback host can bill above it for a turn. source
Aug 26, 2025
Hermes 4 405BLaunchedsecondary sourcein$1.00out$3.00
OpenRouter listing "Nous: Hermes 4 405B" (131K context; one host, Nebius Token Factory, fp8); rate read on the model and endpoint pages 2026-09-23. Llama 3.1 405B base, optional reasoning. source
Jul 9, 2025
Venice UncensoredLaunchedsecondary sourcein$0.20out$0.90
OpenRouter listing "Venice: Uncensored" (128K context, 8K max output; one host, Venice); rate read on the model and endpoint pages 2026-09-23. A Dolphin fine-tune of Mistral-Small-24B-Instruct-2501 by dphn.ai with Venice.ai. source
Apr 5, 2025
Llama 4 MaverickLaunchedsecondary sourcein$0.188out$0.653
OpenRouter listing (1M context); rate read on the model page 2026-09-17; the rate PolyCog estimates with — OpenRouter routes cheapest-host-first with fallbacks up the price ladder (proxy.php, PC84), so a fallback host can bill above it for a turn. Meta's newest text model on OpenRouter as of 2026-09-17 (no Llama 5 listed). source
No dated history for this selection yet.
Since launch
OpenRouter API since May 2023 · first public price Apr 2025Priciest: Llama 4 Maverick at $0.188 / $0.653 → Muse Wonder at $4.00 / $20.00 today. Cheapest: Llama 4 Maverick at $0.188 / $0.653 → Qwen3.8 Flash at $0.15 / $0.47.
Priciest — the most expensive of the seat’s models at the timeCheapest text modelblended 3 in : 1 out · USD per 1M tokens · log scale
Hover or tap for the exact input / output at any change · ← → on a focused chart · 9 dated changes
All 17 OpenRouter models since Apr 2025 — launch and last price
| Model | Tier | Launched | Launch in / out | Last in / out | Retired |
|---|---|---|---|---|---|
| Llama 4 Maverickofferedsecondaryprofile →meta-llama/llama-4-maverickApr 5, 2025 | mid | Apr 5, 2025 | $0.188 / $0.653 | $0.188 / $0.653 | — |
| Venice Uncensoredofferedsecondaryprofile →cognitivecomputations/dolphin-mistral-24b-venice-editionJul 9, 2025 | small | Jul 9, 2025 | $0.20 / $0.90 | $0.20 / $0.90 | — |
| Hermes 4 405Bofferedsecondaryprofile →nousresearch/hermes-4-405bAug 26, 2025 | mid | Aug 26, 2025 | $1.00 / $3.00 | $1.00 / $3.00 | — |
| Mistral Small 4offeredsecondaryprofile →mistralai/mistral-small-2603Mar 16, 2026 | small | Mar 16, 2026 | $0.15 / $0.60 | $0.15 / $0.60 | — |
| Kimi K2.6offeredsecondaryprofile →moonshotai/kimi-k2.6Apr 20, 2026 | mid | Apr 20, 2026 | $0.95 / $4.00 | $0.95 / $4.00 | — |
| Mistral Medium 3.5offeredsecondaryprofile →mistralai/mistral-medium-3-5Apr 30, 2026 | flagship | Apr 30, 2026 | $1.50 / $7.50 | $1.50 / $7.50 | — |
| MiniMax M3offeredsecondaryprofile →minimax/minimax-m3May 31, 2026 | mid | May 31, 2026 | $0.30 / $1.20 | $0.30 / $1.20 | — |
| Kimi K3offeredsecondaryprofile →moonshotai/kimi-k3Jul 16, 2026 | flagship | Jul 16, 2026 | $3.00 / $15.00 | $3.00 / $15.00 | — |
| GLM 5.3offeredsecondaryprofile →z-ai/glm-5.3Aug 18, 2026 | flagship | Aug 18, 2026 | $1.40 / $4.40 | $1.40 / $4.40 | — |
| Qwen3.8 Flashofferedsecondaryprofile →qwen/qwen3.8-flashAug 26, 2026 | small | Aug 26, 2026 | $0.15 / $0.47 | $0.15 / $0.47 | — |
| GLM 5.3 Flashofferedsecondaryprofile →z-ai/glm-5.3-flashAug 26, 2026 | small | Aug 26, 2026 | $0.15 / $0.50 | $0.15 / $0.50 | — |
| Muse Spark 1.3offeredprofile →meta/muse-spark-1.3Sep 2, 2026 | mid | Sep 2, 2026 | $1.25 / $4.25 | $1.25 / $4.25 | — |
| Qwen3.8 Maxofferedsecondaryprofile →qwen/qwen3.8-max-0902Sep 3, 2026 | flagship | Sep 3, 2026 | $2.00 / $6.00 | $2.00 / $6.00 | — |
| Mistral Large 4offeredprofile →mistralai/mistral-large-4-0Oct 6, 2026 | flagship | Oct 6, 2026 | $1.36 / $4.18 | $0.70 / $2.10▼ 49% in ▼ 50% out | — |
| Muse Wonderofferedprofile →meta/muse-wonderOct 12, 2026 | flagship | Oct 12, 2026 | $4.00 / $20.00 | $4.00 / $20.00 | — |
| MiniMax M3.1 Flashofferedprofile →minimax/minimax-m3.1-flashOct 19, 2026 | small | Oct 19, 2026 | $0.15 / $0.60 | $0.15 / $0.60 | — |
| Kimi K3.1 Proofferedprofile →moonshotai/kimi-k3.1-proOct 23, 2026 | flagship | Oct 23, 2026 | $2.00 / $10.00 | $2.00 / $10.00 | — |
Picture prices (Create)
USD per picture on your own key, the figure PolyCog plans with (it rounds up). Draft, Standard and High map to each provider's own settings. Claude and DeepSeek sketch in SVG and pay only their normal text price. When the provider reports the real charge (OpenRouter does), your spend shows that instead.
| Seat | Picture model | Draft | Standard | High |
|---|---|---|---|---|
| ChatGPT | GPT Image 2.5 Flare | $0.006 | $0.013 | $0.053 |
| ChatGPT | GPT Image 2.5 Sunburst | $0.006 | $0.013 | $0.053 |
| ChatGPT | GPT Image 2 | $0.006 | $0.053 | $0.21 |
| Gemini | Nano Banana 2.1 | $0.034 | $0.034 | $0.050 |
| Gemini | Nano Banana 2 | $0.045 | $0.067 | $0.10 |
| Gemini | Nano Banana 2 Lite | $0.034 | $0.034 | $0.034 |
| Gemini | Nano Banana Pro | $0.13 | $0.13 | $0.13 |
| Grok | Grok Imagine 2.0 | $0.040 | $0.060 | $0.080 |
| Grok | Grok Imagine 1.0 | $0.020 | $0.020 | $0.020 |
| Wildcard | FLUX.3 Image | $0.041 | $0.041 | $0.16 |
| Wildcard | Muse Image | $0.010 | $0.010 | $0.040 |
| Wildcard | Qwen Image 3 | $0.033 | $0.033 | $0.13 |
ChatGPT adds about 30% for a wide or tall picture plus two small previews (about $0.006). Gemini bills its pictures as tokens; the figure shown is Google's list price per picture at 512, 1K and 2K (Nano Banana 2.1 has no 512, so its Draft is a 1K picture). The Wildcard's figures are OpenRouter's listed starting price at 1K and four times that at 2K.
What PolyCog itself costs
The prices above are billed by the providers to your own API keys. PolyCog adds nothing to them — it charges for the interface only.
Samplefree
No signup, no keys. Runs on our shared keys.
- 3 prompts a day
- Three fast models (Claude Haiku 5.5, Gemini 2.5 Flash-Lite, DeepSeek V4.1 Flash)
- See the debate format first
Free$0
Bring your own keys. Free forever.
- All six providers, every model above
- 3 questions a day on our small models while you add your keys
- 20 saved sessions
- Up to 3 debate rounds
Paid$10/mo · $250 once
For daily use.
- 500 saved sessions
- Unlimited debate rounds
- Same BYOK billing — no markup, ever
- Cancel in two clicks, anytime; Pro runs to the end of the month you paid for
Common questions
Does PolyCog mark up API prices?
No. PolyCog is BYOK — bring your own key. Your prompts run under your own OpenAI, Anthropic, Google, DeepSeek or xAI account at the provider’s published price, and the charges appear in your own provider dashboard. PolyCog charges for the interface only: free, $10 a month, or $250 once.
Why are some prices red and some green?
Red marks a price that went up on that date (or is announced to go up); green marks a price that went down. Grey means no change. The arrows and percentages say the same thing for anyone reading without colour.
What is a promotional or introductory price?
A cut with a published end date. Right now: GPT-5.6 Sol is $4 / $20 “at least through November 21, 2026” (list $5 / $30), Gemini 4 Argon is $2 / $10 through November 30, 2026, then $4 / $20, and Gemini 3.6, 3.7 and 3.8 Flash are $0.75 / $3.75 through December 31, 2026, then $1.50 / $7.50. The page shows the scheduled change as a dashed red line.
What do DeepSeek’s peak and off-peak rates mean?
Since August 16, 2026 DeepSeek bills a peak rate on weekdays 09:00–12:00 and 14:00–18:00 Beijing time (01:00–04:00 and 06:00–10:00 UTC) and half that at all other times, weekends included. PolyCog’s in-app estimate uses the peak rate so it never under-estimates; your DeepSeek bill reflects the actual window.
What does one PolyCog debate cost?
Cents. Four everyday models with a debate round and a flagship synthesis is roughly ten cents at these prices; every seat a flagship is about forty. The comparison page “AI subscriptions vs BYOK” works through the arithmetic.
How often is this page updated?
Every time PolyCog’s model catalog is refreshed — the “verified” date on each provider section is the day that provider’s pricing page was last re-read. If you spot a change we missed, send a correction and we will fix it and note the change here.
Where do these numbers come from?
The provider pricing pages, linked from each section, and their changelogs and release notes for the dates; the Wildcard seat's rates are OpenRouter's model listings, which publish no history. Where a date is only approximately known the page says so with a ≈ mark.
How far back does the history go?
To each provider’s first public per-token API price: OpenAI’s GPT-3 list of September 2021 (Davinci at $60 per million tokens in and out), Anthropic’s Claude 1 in March 2023, Google’s Gemini 1.0 Pro in February 2024 (originally priced per character — converted at four characters per token and marked ≈), DeepSeek V2 in May 2024 and xAI’s grok-beta in October 2024. Every retired model is in the “every model since …” table under its provider, and the “Retired models” chip adds their price changes to the timeline.
See what these prices buy
Run one question through several frontier models, watch them debate, read the synthesis — and see the cost estimate on every session.
Free tier forever · $10/mo · $250 lifetime · your API keys, your bill
Prices re-verified per provider on the dates shown. Product names are trademarks of their owners; PolyCog is not affiliated with any provider listed. Spotted a change we missed? Send a correction and we’ll note it here.
Data: /api/pricing.json · /api/pricing-history.json · Which model for what? → · All comparisons →