LLM pricing comparison
Every model Vigil's proxy can price, in one table: input, output and cache-read rates per million tokens across 14 platforms, sortable by any column, each row with the date it was verified against the provider's page. Built from the registry on 2026-10-10. Where no verified rate exists, a dash.
72 of 72 rows. USD per 1M tokens, standard tier, global endpoint. Click a column to sort; rows with no figure sort last.
| Rate | |||||
|---|---|---|---|---|---|
| openai | gpt-6-luna | $0.10 | $0.50 | $0.010 | verified 2026-09-30 |
| zai | glm-5.3-flash | $0.15 | $0.50 | $0.030 | verified 2026-09-11 |
| openai | gpt-4o-mini | $0.15 | $0.60 | $0.075 | verified 2026-09-30 |
| groq | openai/gpt-oss-120b | $0.15 | $0.60 | — | verified 2026-09-11 |
| together | zai-org/GLM-5.3-Flash | $0.15 | $0.50 | $0.030 | verified 2026-09-11 |
| openai | gpt-5.6-luna | $0.20 | $1.20 | $0.020 | verified 2026-09-11 |
| deepseek | deepseek-flash Standard price, charged 01:00–04:00 and 06:00–10:00 UTC on weekdays. At every other hour the price is lower: $0.15 input, $0.60 output, $0.003 cache read per million tokens. | $0.30 | $1.20 | $0.006 | verified 2026-09-11 |
| openai | gpt-4.1-mini | $0.40 | $1.60 | $0.10 | verified 2026-09-30 |
| gemini-3.6-flash | $0.75 | $3.75 | $0.075 | verified 2026-09-11 | |
| gemini-3.7-flash | $0.75 | $3.75 | $0.075 | verified 2026-09-11 | |
| gemini-3.8-flash | $0.75 | $3.75 | $0.075 | verified 2026-09-11 | |
| anthropic | claude-3-5-haiku | $0.80 | $4.00 | $0.080 | verified 2026-09-30 |
| groq | Qwen/Qwen3.8-27B | $0.80 | $4.00 | — | verified 2026-09-11 |
| moonshot | kimi-k2.6 | $0.95 | $4.00 | $0.16 | verified 2026-09-11 |
| anthropic | claude-haiku-4-5-20251001 | $1.00 | $5.00 | $0.10 | verified 2026-09-11 |
| vertex | claude-haiku-4-5@20251001 | $1.00 | $5.00 | $0.10 | verified 2026-09-11 |
| xai | grok-build-0.1 | $1.00 | $2.00 | $0.20 | verified 2026-09-11 |
| together | meta-llama/Llama-3.3-70B-Instruct-Turbo | $1.04 | $1.04 | — | verified 2026-09-11 |
| xai | grok-4.3 | $1.25 | $2.50 | $0.20 | verified 2026-09-11 |
| together | deepseek-ai/DeepSeek-V4-Pro-0813 | $1.32 | $3.96 | $0.13 | verified 2026-09-11 |
| deepseek | deepseek-v4-pro | $1.32 | $3.96 | $0.044 | verified 2026-09-11 |
| fireworks | fireworks/deepseek-v4-pro-0813 | $1.32 | $3.96 | $0.044 | verified 2026-09-11 |
| fireworks | fireworks/glm-5p3 | $1.40 | $4.40 | $0.26 | verified 2026-09-11 |
| zai | glm-5.3 | $1.40 | $4.40 | $0.26 | verified 2026-09-11 |
| baseten | zai-org/GLM-5.3 | $1.40 | $4.40 | $0.14 | verified 2026-09-11 |
| gemini-3.5-flash | $1.50 | $9.00 | $0.15 | verified 2026-09-30 | |
| openai | gpt-5.3-codex | $1.75 | $14.00 | $0.17 | verified 2026-09-11 |
| anthropic | claude-sonnet-5 | $2.00 | $10.00 | $0.20 | verified 2026-09-11 |
| vertex | claude-sonnet-5 | $2.00 | $10.00 | $0.20 | verified 2026-09-11 |
| gemini-3.1-pro-preview | $2.00 | $12.00 | $0.20 | verified 2026-09-30 | |
| openai | gpt-5.6-terra | $2.00 | $12.00 | $0.20 | verified 2026-09-11 |
| openai | gpt-6-sol | $2.00 | $10.00 | $0.20 | verified 2026-09-30 |
| xai | grok-4.5 | $2.00 | $6.00 | $0.30 | verified 2026-09-11 |
| xai | grok-4.6 | $2.00 | $6.00 | $0.50 | verified 2026-09-11 |
| openai | gpt-4o | $2.50 | $10.00 | $1.25 | verified 2026-09-30 |
| anthropic | claude-sonnet-4 | $3.00 | $15.00 | $0.30 | verified 2026-09-30 |
| anthropic | claude-sonnet-4-5 | $3.00 | $15.00 | $0.30 | verified 2026-09-30 |
| vertex | claude-sonnet-4-5@20250929 | $3.00 | $15.00 | $0.30 | verified 2026-09-11 |
| anthropic | claude-sonnet-4-6 | $3.00 | $15.00 | $0.30 | verified 2026-09-11 |
| vertex | claude-sonnet-4-6 | $3.00 | $15.00 | $0.30 | verified 2026-09-11 |
| fireworks | fireworks/kimi-k3 | $3.00 | $15.00 | $0.30 | verified 2026-09-11 |
| moonshot | kimi-k3 | $3.00 | $15.00 | $0.30 | verified 2026-09-11 |
| baseten | moonshotai/Kimi-K3 | $3.00 | $15.00 | $0.30 | verified 2026-09-11 |
| openai | gpt-5.6-sol | $4.00 | $20.00 | $0.40 | verified 2026-09-11 |
| openai | chat-latest | $5.00 | $30.00 | $0.50 | verified 2026-09-11 |
| anthropic | claude-opus-4-5 | $5.00 | $25.00 | $0.50 | verified 2026-09-30 |
| vertex | claude-opus-4-5@20251101 | $5.00 | $25.00 | $0.50 | verified 2026-09-11 |
| anthropic | claude-opus-4-6 | $5.00 | $25.00 | $0.50 | verified 2026-09-30 |
| vertex | claude-opus-4-6 | $5.00 | $25.00 | $0.50 | verified 2026-09-11 |
| anthropic | claude-opus-4-7 | $5.00 | $25.00 | $0.50 | verified 2026-09-30 |
| vertex | claude-opus-4-7 | $5.00 | $25.00 | $0.50 | verified 2026-09-11 |
| anthropic | claude-opus-4-8 | $5.00 | $25.00 | $0.50 | verified 2026-09-11 |
| vertex | claude-opus-4-8 | $5.00 | $25.00 | $0.50 | verified 2026-09-11 |
| anthropic | claude-opus-5 | $5.00 | $25.00 | $0.50 | verified 2026-09-11 |
| vertex | claude-opus-5 | $5.00 | $25.00 | $0.50 | verified 2026-09-11 |
| openai | gpt-4o-2024-05-13 | $5.00 | $15.00 | — | verified 2026-09-30 |
| anthropic | claude-fable-5 | $10.00 | $50.00 | $1.00 | verified 2026-09-11 |
| vertex | claude-fable-5 | $10.00 | $50.00 | $1.00 | verified 2026-09-11 |
| anthropic | claude-fable-5-1 | $10.00 | $50.00 | $0.25 | verified 2026-09-11 |
| vertex | claude-fable-5-1 | $10.00 | $50.00 | $0.25 | verified 2026-09-11 |
| openai | gpt-6-astra | $10.00 | $50.00 | $1.00 | verified 2026-09-11 |
| openai | gpt-5.6-cyber | $12.50 | $75.00 | $1.25 | verified 2026-09-11 |
| bedrock | anthropic.claude-fable-5 bare Bedrock id 'anthropic.claude-fable-5' is an in-region invocation; the global endpoint is the global. inference profile — endpoint_class 'global' contradicts the id | — | — | — | contract 2026-09-11 |
| bedrock | anthropic.claude-fable-5-1 bare Bedrock id 'anthropic.claude-fable-5-1' is an in-region invocation; the global endpoint is the global. inference profile — endpoint_class 'global' contradicts the id | — | — | — | contract 2026-09-11 |
| bedrock | anthropic.claude-haiku-4-5-20251001 bare Bedrock id 'anthropic.claude-haiku-4-5-20251001' is an in-region invocation; the global endpoint is the global. inference profile — endpoint_class 'global' contradicts the id | — | — | — | contract 2026-09-11 |
| bedrock | anthropic.claude-opus-4-8 bare Bedrock id 'anthropic.claude-opus-4-8' is an in-region invocation; the global endpoint is the global. inference profile — endpoint_class 'global' contradicts the id | — | — | — | verified 2026-09-11 |
| bedrock | anthropic.claude-opus-5 bare Bedrock id 'anthropic.claude-opus-5' is an in-region invocation; the global endpoint is the global. inference profile — endpoint_class 'global' contradicts the id | — | — | — | contract 2026-09-11 |
| bedrock | anthropic.claude-sonnet-4-5-20250929 bare Bedrock id 'anthropic.claude-sonnet-4-5-20250929' is an in-region invocation; the global endpoint is the global. inference profile — endpoint_class 'global' contradicts the id | — | — | — | verified 2026-09-11 |
| bedrock | anthropic.claude-sonnet-4-6 bare Bedrock id 'anthropic.claude-sonnet-4-6' is an in-region invocation; the global endpoint is the global. inference profile — endpoint_class 'global' contradicts the id | — | — | — | verified 2026-09-11 |
| bedrock | anthropic.claude-sonnet-5 bare Bedrock id 'anthropic.claude-sonnet-5' is an in-region invocation; the global endpoint is the global. inference profile — endpoint_class 'global' contradicts the id | — | — | — | contract 2026-09-11 |
| cerebras | gpt-oss-120b rate for (cerebras, gpt-oss-120b) is unverified — cerebras.ai pricing table JS-rendered, not retrievable this pass (checked 2026-09-11) — an unverified rate is never substituted with a plausible value | — | — | — | unverified 2026-09-11 |
| bedrock | meta.llama3-3-70b-instruct bare Bedrock id 'meta.llama3-3-70b-instruct' is an in-region invocation; the global endpoint is the global. inference profile — endpoint_class 'global' contradicts the id | — | — | — | unverified 2026-09-11 |
Reading the table
- Prices are the standard, on-demand tier on the global endpoint class. Batch, regional and long-context rates are on each model's own page, linked from its name where one exists.
- A model billed by UTC hour (DeepSeek) shows its standard price with a note giving the off-peak one.
- “Verified” means the row was read against the provider's published price on the date shown. “Contract” means the provider's page names that platform's price as the contractual rate. “Inferred” means the proxy derived it from a stated rule and no published row.
The same model id is not the same price everywhere, and a price per token is only part of a bill: the posts below cover cache writes, hidden costs and what tool definitions cost on every call.
Questions
- Where do these prices come from?
- From the rate registry Vigil’s proxy bills real traffic with, generated into this site at build time. Every row names the provider page it was read from and the date; a row’s label says whether the rate was verified against that page, designated by contract, or inferred.
- Why do some rows show a dash?
- Because no verified rate exists for that model, and a plausible guess is worse than an honest gap. A dash sorts last in every direction so it can never look like the cheapest option.
- Why is the same model listed on more than one platform?
- Because it is a different product on each: Claude on Anthropic, on AWS Bedrock and on Google Vertex has separately published rates, different endpoint classes and three separate prompt caches. The table keeps them apart so a comparison compares like with like.