AIIntelGet the Briefing

Latest

Models

Get the Briefing

AI-drafted from cited sources and checked against our publishing rules. Not every change is reviewed by a person before it goes live.

Leading AI models, one row per model.

Official figures only, as of 30 October 2026. Nothing here is estimated.

Models
22

Each row links to its source. A dash means not yet reported.

Leading AI models. 22 shown.
ModelMakerMaker countryTypeRelease dateContext window (tokens)Input $/1MOutput $/1MOpen weightsAccessStatusNotesSourceLast checked
Mistral Large 4Mistral—multimodal29 October 20261M tokens$0.70$2.10Yesopen weights and APIreleasedWeights released 29 October 2026. Public preview was 6 October 2026. Cached input is $0.07 per million tokens. Hugging Face card: 49 billion active parameters per token, 52 billion including embeddings and output layers. Technical report not yet published. Mistral Ecosystem Open-Weight licence; a model-as-a-service offering needs a separate commercial licence.
Gemini 4 CarbonGoogle——28 October 2026—$2 introductory, then $4$10 introductory, then $20—Limited to cyber defenders, then the Fairwind ProgramlimitedCached input is priced at 95% off the input token price. The post does not say when the introductory period ends and does not state a context window. Not generally available.
Nemotron 3.5 TranscribeNVIDIA—speech28 October 202665,536 tokens——Yesopen weightsreleasedModel card: 34 billion total parameters, 4.3 billion active, including the audio encoder. Recordings up to one hour. Licence OpenMDW-1.1. The repository id says 30B-A4B; the card's total is 34 billion. No token price is stated.
GPT-6.1 TerraOpenAI—multimodal26 October 20261,050,000 tokens$1$5—API, ChatGPT Work, CodexreleasedStandard API price. Cached input is $0.05 per million tokens. The model page says prompts with more than 272K input tokens are priced at 2x input and cache rates and 1.5x output for the full request. Preparedness Framework: Critical in cybersecurity and High for biological and chemical capability. Knowledge cutoff on the model page is Jun 01, 2026.
MiniMax M3.1 FlashMiniMax—LLM19 October 20261M tokens$0.15$0.60YesAPI, open weightsreleasedCached input is $0.015 per million tokens. The maker’s model card publishes the weights.
GlimpseElorian—visual thinking16 October 2026—$0.60$3.00—API, apppreviewResearch preview. The page says Glimpse was post-trained on an open-weights base model. It does not say Glimpse’s own weights are open.
Muse Spark OpenMetaUnited Statesmultimodal15 October 20261M tokens——Yesopen weightsreleased—
Gemini 4 ArgonGoogle——13 October 20261M tokens, input and output together$2 until 30 November 2026, then $4$10 until 30 November 2026, then $20—APIreleased—
Claude Haiku 5.5Anthropic—LLM7 October 20261M tokens$0.10 up to 100k prompt tokens; $0.50 above that$0.50 up to 100k prompt tokens; $2.50 above that—APIreleasedThe 100k split is a price break. The models overview lists a 1M context window. Haiku 5.5 is outside the standard 1M price on the pricing page.
GPT-6.1 SolOpenAI—multimodal29 September 20261,050,000 tokens$2$10—APIreleasedThe model page says input can be text or image, and output is text.
Claude Sonnet 5.5Anthropic—LLM28 September 20261M tokens$2$10—APIreleased—
Claude Opus 5.5Anthropic—LLM22 September 20261M tokens$4$20—APIreleased—
GPT-6 LunaOpenAI—multimodal22 September 20261,050,000 tokens$0.10$0.50—APIreleasedThe model page says input can be text or image, and output is text.
GPT-6 SolOpenAI——22 September 20261,050,000 tokens$2$10—APIreleasedGPT-6.1 Sol is the newer Sol model.
Grok 4.7SpaceXSI (xAI)——21 September 2026500k tokens$2 under 200k prompt tokens; $4 at 200k or above$6 under 200k prompt tokens; $12 at 200k or above—APIreleasedA prompt at or above 200k tokens is billed at the higher rate for every token in the request.
DeepSeek-V4.1-FlashDeepSeek——10 September 20261M tokensCache miss $0.15 off-peak or $0.30 peak$0.60 off-peak or $1.20 peakYesAPI, open weightsreleasedThe news post says it is live and has native visual understanding. Peak hours are 01:00–04:00 and 06:00–10:00 UTC, Monday to Friday, excluding Chinese public holidays. The maker’s model card licenses the weights under MIT.
Gemini 3.8 FlashGoogle——2 September 2026—$0.75 until 31 December 2026, then $1.50$3.75 until 31 December 2026, then $7.50—APIreleasedIntroductory price, described as the same as 3.7 Flash.
Claude Fable 5.1Anthropic—LLMSeptember 20261M tokens$10$50—APIreleasedThe announcement gives the month, not the day.
Claude Mythos 5.1Anthropic—LLMSeptember 20261M tokens$10$50—APIreleasedThe announcement gives the month, not the day. Limited availability on the pricing page.
DeepSeek-V4-ProDeepSeek—LLM—1M tokensCache miss $0.66 off-peak or $1.32 peak$1.98 off-peak or $3.96 peakYesAPI, open weightspreviewThe maker’s model card calls this a preview. The same peak hours as Flash apply. The version string is not treated as a release date.
GPT-6 AstraOpenAI———1,050,000 tokens$10$50—APIreleasedA prompt over 272,000 input tokens is billed at twice the input rate and 1.5 times the output rate for the whole request. The model page does not give a release date.
Qwen3.8-MaxAlibaba Cloud———Up to 1M input tokens$2 (Singapore, international)$6 (Singapore, international)—API—Model Studio pricing page, last updated 22 October 2026. Other regions on the same page are priced differently.