Skip to content

Model reference: OpenAI and Claude

A side-by-side of the models you can point this Action at, as of 2026-09-20. It exists so you can compare options when choosing openai_model or anthropic_model without opening a dozen provider pages.

⚠️ Important disclaimer: read this first

This page is provided for convenience only. Do not rely on it for pricing, budgeting, or purchasing decisions.

  • The figures may be wrong for you. They were gathered on 2026-09-20 from provider documentation, including pages viewed from a personal account, and may not reflect the prices, limits, or model availability that apply to your account, tier, region, or contract.
  • Your costs may differ. Real cost depends on your usage tier, data-residency or regional surcharges, batch, flex, or fast processing, promotional or negotiated rates, taxes, and how many tokens your pull requests actually use.
  • It goes out of date. Providers change prices, add models, and retire them, often with little notice. Nothing here is updated automatically.
  • It is not a quote, a guarantee, or an endorsement. The providers' own pages are the only authoritative source (see Sources). Always check them before you rely on a number.
  • Being listed here does not mean the Action supports a model. The Action's defaults are gpt-5-mini and claude-haiku-4-5-20251001; other models are not guaranteed to work.

This project is not affiliated with OpenAI or Anthropic. Product and model names belong to their owners.

All prices are US dollars per million tokens (MTok), for the standard tier, "input / output" unless stated. Context, output, and cutoff figures are as the providers state them.

OpenAI

How it compares

Model Context Max input Max output Price / MTok (in / out) Thinking Default effort Knowledge cutoff
GPT-6 Astra 1.05M 922K 128K $10 / $50 Reasoning; effort low to max (no none) Not stated on the model page Apr 30, 2026
GPT-5.6 Sol 1.05M 922K 128K $4 / $20 (promotional pricing through at least Nov 21, 2026) Reasoning; effort none to max medium Feb 16, 2026
GPT-5.6 Terra 1.05M 922K 128K $2 / $12 Reasoning; effort none to max medium Feb 16, 2026
GPT-5.6 Luna 1.05M 922K 128K $0.20 / $1.20 Reasoning; effort none to max medium Feb 16, 2026
GPT-5 Mini (the Action's default) 400K 272K 128K $0.25 / $2 Reasoning Not stated on the model page May 31, 2024

The OpenAI pages used here give no latency rating. OpenAI offers Fast mode instead (about 2x the price for up to 2.5x the speed). All of these take text and images in and give text out.

Model IDs, pricing detail and availability

Model Model ID Cached input Cache write Batch / Flex (50% off) Released Retirement
GPT-6 Astra gpt-6-astra $1 $12.50 $5 / $25 Sep 3, 2026 None announced
GPT-5.6 Sol gpt-5.6-sol (alias gpt-5.6) $0.40 $5 $2 / $10 Jul 9, 2026 None announced
GPT-5.6 Terra gpt-5.6-terra $0.20 $2.50 $1 / $6 Jul 9, 2026 None announced
GPT-5.6 Luna gpt-5.6-luna $0.02 $0.25 $0.10 / $0.60 Jul 9, 2026 None announced
GPT-5 Mini gpt-5-mini (snapshot gpt-5-mini-2025-08-07) $0.025 Not listed Not listed Aug 2025 (snapshot date) Snapshot listed for removal on Dec 11, 2026
  • Prompts over 272K input tokens are priced at 2x input and 1.5x output for the whole request (GPT-6 Astra and the GPT-5.6 models).
  • Cache writes are billed at 1.25x the uncached input rate.
  • Regional processing (data residency) endpoints carry a 10% uplift for eligible models released on or after March 5, 2026.
  • OpenAI describes Sol, Terra and Luna as roughly the standard, mini and nano tiers of earlier GPT-5 families.
  • OpenAI states that GPT-6 Astra does not support custom temperature, top_p or log probabilities. The Action treats gpt-6* models like the other reasoning models, so it never sends temperature to them and uses max_completion_tokens.

Claude (Anthropic)

How it compares

Model Context Max output Price / MTok (in / out) Latency Thinking Default effort Knowledge cutoff
Claude Fable 5.1 1M 128K $10 / $50 Slower Adaptive (always on) high Jun 2026
Claude Opus 5 1M 128K $5 / $25 Moderate Adaptive high May 2026
Claude Sonnet 5 1M 128K $2 / $10 Fast Adaptive high Jan 2026
Claude Haiku 4.5 (the Action's default) 200K 64K $1 / $5 Fastest Extended Not supported Feb 2025

All of these take text and images in and give text out. Only the comparison row was captured for Claude Fable 5.1, so it has no detail row below.

Model IDs, pricing detail and availability

Model Model ID (Claude API) Cache write (5 min / 1 hour) Cache read Batch Released Retirement
Claude Opus 5 claude-opus-5 $6.25 / $10 $0.50 50% off input and output Jul 24, 2026 Not sooner than Jul 24, 2027
Claude Sonnet 5 claude-sonnet-5 $2.50 / $4 $0.20 50% off input and output Jun 30, 2026 Not sooner than Jun 30, 2027
Claude Haiku 4.5 claude-haiku-4-5-20251001 (alias claude-haiku-4-5) $1.25 / $2 $0.10 50% off input and output Oct 15, 2025 Not sooner than Oct 15, 2026
  • Sonnet 5 and Opus 5 list a larger maximum output for the Batch API (300K tokens, beta).
  • Haiku 4.5's other platform IDs: Amazon Bedrock anthropic.claude-haiku-4-5, Google Cloud claude-haiku-4-5@20251001, Microsoft Foundry claude-haiku-4-5. This Action uses the direct Claude API only.
  • Haiku 4.5 lists a reliable knowledge cutoff of Feb 2025 and a training-data cutoff of Jul 2025; Sonnet 5 lists Jan 2026 for both, and Opus 5 lists May 2026 for both.
  • Sonnet 5 and Opus 5 think by default, and that counts against max_tokens. See Choosing a provider.

Retirement dates for the defaults

Both default models have a date worth knowing about. These come from the same sources and carry the same disclaimer:

  • gpt-5-mini: OpenAI's deprecations page lists its snapshot, gpt-5-mini-2025-08-07, for removal on Dec 11, 2026, with gpt-5.6-terra as the suggested replacement.
  • claude-haiku-4-5-20251001: Anthropic lists its retirement as not sooner than Oct 15, 2026. That is a floor, not a scheduled shutdown date.

If a default is retired, set openai_model or anthropic_model explicitly.

Sources

Compiled on 2026-09-20 from:

  • OpenAI: the models, pricing, deprecations and changelog pages, and the individual model pages for gpt-6-astra, gpt-5.6-sol, gpt-5.6-terra, gpt-5.6-luna and gpt-5-mini.
  • Anthropic: the Claude models overview and the individual model pages for Haiku 4.5, Sonnet 5 and Opus 5, transcribed from screenshots taken from a personal account.

Reminder: check the provider's own pages before relying on any figure above.