LIVE
Anthropic says Claude helped trace the origin of NIST's mysterious ECDSA seeds16/10/26 · Anthropic|Reflection AI unveils Beam, a 501-billion-parameter open model16/10/26 · Reflection AI|Meta releases Muse Spark Open, a mixture-of-experts model built for local Mac Studio inference15/10/26 · Meta|Meta releases Muse Spark Open, an open-weight multimodal model15/10/26 · Meta|Google DeepMind outlines safety measures behind Gemini 4 Argon13/10/26 · Google DeepMind|Tev-Gym: A Training Recipe to Fix Sparse Rewards in Long-Horizon Agents13/10/26|Google DeepMind references Gemini 4, code-named Argon13/10/26 · Google DeepMind|Google DeepMind unveils Gemini 4 Argon, a robotics-focused variant13/10/26 · Google DeepMind|Character.AI files for Chapter 11 bankruptcy, sells assets to Disney13/10/26 · Character.AI|Study Shows AI Models Can Provably Conceal Arbitrary Capabilities13/10/26|Mistral AI releases Timestral, a foundation model for time-series data19/10/26 · Mistral AI|Researchers strand Claude on a virtual desert island to study unscripted behavior19/10/26 · Anthropic|Anthropic says Claude helped trace the origin of NIST's mysterious ECDSA seeds16/10/26 · Anthropic|Reflection AI unveils Beam, a 501-billion-parameter open model16/10/26 · Reflection AI|Meta releases Muse Spark Open, a mixture-of-experts model built for local Mac Studio inference15/10/26 · Meta|Meta releases Muse Spark Open, an open-weight multimodal model15/10/26 · Meta|Google DeepMind outlines safety measures behind Gemini 4 Argon13/10/26 · Google DeepMind|Tev-Gym: A Training Recipe to Fix Sparse Rewards in Long-Horizon Agents13/10/26|Google DeepMind references Gemini 4, code-named Argon13/10/26 · Google DeepMind|Google DeepMind unveils Gemini 4 Argon, a robotics-focused variant13/10/26 · Google DeepMind|Character.AI files for Chapter 11 bankruptcy, sells assets to Disney13/10/26 · Character.AI|Study Shows AI Models Can Provably Conceal Arbitrary Capabilities13/10/26|Mistral AI releases Timestral, a foundation model for time-series data19/10/26 · Mistral AI|Researchers strand Claude on a virtual desert island to study unscripted behavior19/10/26 · Anthropic|
≤ 2 $

Best API LLMs under $2 per million output tokens

API-accessible models with an output price at or under $2 per million tokens, ranked by Elo. The real quality-to-cost ratio for high-volume pipelines — classification, extraction, summarisation — where price per token matters more than the last benchmark point.

#ModelArena EloOutput / M tokensHostingLicence
1🇨🇳Qwen3.7 Plus
Alibaba
1456$1.60/MAPIproprietary
2🇺🇸GPT-5.6 Luna
OpenAI
1453$1.20/MEU optionproprietary
3🇺🇸GPT-6 Luna
OpenAI
1443$0.50/MEU optionproprietary
4🇫🇷Mistral Large 3
Mistral AI
1414$1.50/MSelf-hostableapache-2.0
5🇺🇸GPT-5 mini
OpenAI
1390$2.00/MEU optionproprietary
6🇫🇷Codestral
Mistral AI
1240$0.90/MSelf-hostablemistral-non-commercial
7🇫🇷Mistral Small 4
Mistral AI
—$0.60/MSelf-hostableapache-2.0
8🌍Claude Haiku 5.5
Anthropic
—$0.50/MAPI—
Data verified on 19 October 2026Open in the comparator

Method

Deterministic ranking on comparator data: LMArena Elo (style control) first, HuggingFace downloads second for models without an Elo. Prices and licences are verified at the official source and dated. No model is ranked by an LLM.

Other guides