Current AI Models: Which One to Pick for Each Task

Comparison guide to 12 AI models as of November 15, 2026: release date, context, price, open or closed weights and which one to try first for your task.

info Quick Definition
This guide compares 12 AI models from nine labs (Anthropic, OpenAI, Google, xAI, Meta, DeepSeek, Alibaba, Mistral and Moonshot) using data from their official pages, and suggests which to try first for each task. Data is as of November 15, 2026, the cutoff date; prices and licenses change.

Which models are the most recent?

This sheet covers the 12 models that have their own page in the wiki. Prices are USD per million tokens, standard input and output rates.

ModelLabDateContextInput / outputWeights
Claude Fable 5.2AnthropicNov 9, 20261M8 / 40Closed
Claude Opus 5.5AnthropicSep 22, 20261M4 / 20Closed
Claude Haiku 5.5AnthropicOct 7, 20261M0.10 / 0.50 (0.50 / 2.50 above 100,000 prompt tokens)Closed
GPT-6 AstraOpenAISep 3, 2026 (system card)1,050,00010 / 50 (up to 272K input)Closed
Gemini 4 CarbonGoogleNov 10, 2026 on the API1M2 / 10 until Nov 30; 4 / 20 from Dec 1Closed
Gemini 4 HydrogenGoogleNov 12, 20261M1.50 / 7.50Closed
Grok 4.7xAISep 21, 2026500K2 / 6 (4 / 12 from 200K)xAI does not announce downloadable weights
Muse EmberMetaNov 9, 20261M0.10 / 0.50 (0.30 / 1.50 above 200K input)Closed
DeepSeek V4.1 FlashDeepSeekSep 10, 2026 (version 1017: Oct 17)1M0.15 / 0.60 off-peak (0.30 / 1.20 at peak), uncached inputOpen, MIT
Qwen3.8-MaxAlibabaAug 3, 20261M2 / 6 (Singapore and US)Weights under its own license
Mistral Large 4MistralOct 6, 2026 (preview)1M0.70 / 2.10Open weights, Mistral license
Kimi K3.1 ProMoonshotnot confirmed (weights by Nov 4, 2026 at the latest)1Mnot confirmedOpen weights, its own license

How do they compare on price and context?

  • Context: almost all list 1M tokens. GPT-6 Astra reaches 1,050,000 and Grok 4.7 has 500K.
  • Cheapest: Haiku 5.5 and Muse Ember, both positioned by their labs for repetitive, high-volume work. See small vs. large models.
  • Most expensive: Fable 5.2 (8 / 40) and GPT-6 Astra (10 / 50), presented for the most demanding work.
  • Price depends on prompt length: Haiku 5.5, GPT-6 Astra, Grok 4.7 and Muse Ember charge more once the prompt passes a threshold, and it applies to the whole request. Gemini 4 Hydrogen says it keeps a flat price across its full million tokens.
  • Price depends on date or time of day: Carbon goes up on December 1 and DeepSeek charges half off-peak.

To turn this into a real bill, read cost per token.

Which model fits each task?

Each row is a wiki suggestion to try first, backed by a fact from its page. It is not a quality ranking.

If you need…Start with…Why (fact from the wiki)
To classify or extract from thousands of textsHaiku 5.5 or Muse Ember0.10 / 0.50 and positioned for high volume
A default model for code and long agentsOpus 5.5Anthropic recommends it as the starting point; 4 / 20
More power if Opus 5.5 falls shortFable 5.2Anthropic suggests it after trying Opus at high effort
A lot of material in one requestHydrogen or AstraHydrogen: flat price across 1M; Astra: 1.05M, with a surcharge above 272K
Code inside CursorGrok 4.7Available in Cursor; 2 / 6 under 200K prompts
Cybersecurity or long engineering workCarbonGoogle positions it there; price rises on December 1
Fast everyday agentsHydrogenGoogle calls it its daily model; 1.50 / 7.50
Weights under the MIT licenseDeepSeek V4.1 FlashWeights on Hugging Face; 0.15 / 0.60 off-peak
Open weights with a published priceMistral Large 40.70 / 2.10 and 1M context
Code with weights and image inputQwen3.8-Max or Kimi K3.1 ProQwen: 2 / 6; Kimi: its price is not in its announcement or documentation

Which are open and which are closed?

  • Closed, API or product only: the three Claude models on the list, GPT-6 Astra, both Gemini 4 models and Muse Ember. For Grok 4.7, xAI does not announce downloadable weights.
  • With downloadable weights: DeepSeek V4.1 Flash (MIT), Qwen3.8-Max (its own license), Mistral Large 4 and Kimi K3.1 Pro (its own license).

Open does not mean unconditional. Qwen and Kimi ask for a separate agreement if you offer the model as a service to third parties at high revenue (Qwen: over 50 million dollars in 12 months; Kimi: 20 million). Mistral Large 4 has two license descriptions (modified MIT and Apache 2.0): ask legal to review the actual file. More in open vs. closed models.

Why can’t I compare their benchmarks directly?

Because each lab measures with its own harness, its own settings and, sometimes, a different version of the same benchmark. Examples from the pages:

  • The same name gives very different numbers. DeepSeek publishes 35.4 on Agents’ Last Exam; Meta publishes 61.6 for Muse Ember, on a Linux subset it describes as an internal evaluation.
  • Versions differ. Alibaba reports Terminal Bench 2.1; other labs report Terminal-Bench 4.0.
  • Settings change the number: reasoning effort, with or without tools, the harness used.

Use scores to see a model’s trend against its own previous version, in the same table from the same lab. To choose between labs, run your own tests: how to read AI benchmarks.

Example

Illustrative example: a junior designer needs to tag user comments and does not know which model to use. A plan to test three:

  1. Gather 30 real comments and write the correct tag for each by hand.
  2. Pick three: a cheap one (Haiku 5.5), a default one (Opus 5.5) and one from another lab (Hydrogen).
  3. Run the same prompt on all three and note hits, tokens used and cost using this guide’s prices.
  4. Review the misses: did it invent tags or fail on ambiguous cases?
  5. Keep the cheapest one that is accurate enough, and recheck prices in a month.

Common mistakes

  • Choosing by list price without measuring accuracy on your data.
  • Treating first place on a lab’s table as the best for your case.
  • Comparing scores from different labs as if they were the same exam.
  • Missing that a long prompt changes the rate for the whole request.
  • Assuming open means free of conditions.
  • Budgeting with today’s prices and never looking again.

Sources

Frequently asked questions

Which model has the lowest input price?
As of November 15, 2026, Claude Haiku 5.5 and Muse Ember list 0.10 dollars for input and 0.50 for output per million tokens, with a higher rate once the prompt passes 100,000 and 200,000 tokens, respectively. That is list price, not quality: measure with your own data.
How much do the models in this guide cost?
According to each lab’s pages as of November 15, 2026, the standard input price runs from 0.10 dollars per million tokens (Haiku 5.5, Muse Ember) to 10 (GPT-6 Astra), and the output price from 0.50 to 50. Kimi K3.1 Pro’s price is not in its announcement or its models documentation. Prices change: check the official page before budgeting.
Does any price change soon?
Yes. Gemini 4 Carbon costs 2 dollars for input and 10 for output until November 30, 2026, and 4 and 20 from December 1, according to Google AI Studio’s pricing page.
Does this guide say which model is the best?
No. It summarizes data each lab publishes and suggests where to start testing. Benchmarks from different labs cannot be compared with each other, so the final call comes from your own tests.