AI models & prices
68 modelsEvery model below is reachable through one OpenAI-compatible endpoint and a single key. Pay per token — no subscriptions, no per-model accounts.
AnyModel puts the leading models from OpenAI, Anthropic, Google, DeepSeek and others behind a single OpenAI-compatible endpoint. Instead of juggling a separate account, key and bill for each provider, you reach all of them with one key and one balance.
Picking a model comes down to three things: capability, context window and price. Flagships like GPT-6 Astra, Claude Opus 4.8 and Gemini 3.1 Pro handle the hardest reasoning and agentic coding; lighter tiers such as Claude Haiku, GPT-5.6 Luna and Gemini 3.1 Flash-Lite cost a fraction as much and suit high-volume or budget-sensitive work. Context windows run from 200K up to 1M tokens — the larger the window, the more code or documents a model can read in one pass.
The large figure on each card is what you actually pay per 1M tokens — final, with the model's multiplier already applied; metered routes show input and output rates separately. Next to it sits the vendor's own list price and how many times cheaper AnyModel is. You pay only for the tokens you actually use, with no subscription and no per-model minimums.
The newest Opus, at the same price as Opus 5.
cc/claude-opus-5-5OpenAI's most capable model for complex, demanding work.
cx/gpt-6-astraOpenAI's latest GPT-6 workhorse — near-Astra coding at Sol cost.
cx/gpt-6.1-solThe fast, affordable GPT-6 — sub-agents and high-volume work.
cx/gpt-6-lunaNative multimodal coding at flash prices, with a 1M window.
glm/glm-5.3-flashThe newest mid-tier Claude — Sonnet 5 generation, priced to move.
cc/claude-sonnet-5-5Zhipu's newest flagship — the coding-agent GLM.
glm/glm-5.3Latest frontier agentic coding model — the new Codex default.
cx/gpt-5.6-solOpenAI's GPT-6 workhorse for coding and everyday work.
cx/gpt-6-solFast, efficient GPT-6.1 for visual reasoning, computer use and natural writing.
cx/gpt-6.1-terraThe balanced Claude — fast, smart, affordable.
ag/claude-sonnet-4-6The current Opus — Anthropic's agentic-coding flagship.
cc/claude-opus-4-8The newest experimental Opus, in limited availability.
cc/claude-opus-5Balanced agentic coding model for everyday work.
cx/gpt-5.6-terraNVIDIA's open reasoning flagship, available free through AnyModel.
am/nemotron-3-ultra-550b-a55bThe fast, light GPT-5.6 — cheap agents and high-volume coding.
cx/gpt-5.6-lunaCheapest, fastest Claude for high-volume tasks.
cc/claude-haiku-4-5-20251001Automatically selects a currently available free model.
am/freeThe large figure is AnyModel's final price per 1M tokens — the model's multiplier is already in it; input/output rates are shown separately when they differ. The vendor's list price is shown next to it for comparison. USD.
Embeddings and reranking
Vector embeddings for search and RAG through POST /v1/embeddings, and document reranking through POST /v1/rerank — the same key and balance as the text models. Prices are per 1M input tokens.
| Model | Type | Vector size | Max input | Price per 1M tokens |
|---|---|---|---|---|
Nemotron 3.5 Embed 8B am/nemotron-3.5-embed-8b | Embeddings | 4,096shortenable | — | Free |
Nemotron 3 Embed 1B am/nemotron-3-embed-1b | Embeddings | 2,048 | 4,096 tokens | Free |
Nemotron 3.5 Rerank 8B am/nemotron-3.5-rerank-8b | Rerank | — | — | Free |
Max input is shown where we measured it on the live endpoint; an input over the limit is rejected unless you send truncate: END. Vectors of different models are not comparable — build the index and search it with the same model. Parameters and examples are in the API reference.
AnyModel