GitHub Copilot now offers models from OpenAI, Anthropic, Google, xAI, Microsoft and Moonshot AI, and the list changes almost every month. This guide reflects GitHub's documentation as of October 21, 2026 (Beijing time) and will be updated as models are added or retired.

How Copilot charges for models

Copilot usage is billed in GitHub AI Credits, where 1 credit equals $0.01. Each interaction is priced by the tokens it consumes at the chosen model's per-token rate, so a heavier model or a larger context window uses more credits. Copilot Free, Pro, Pro+ and Max include a monthly credit allowance; Copilot Business and Copilot Enterprise include per-user allowances pooled across the organization. Usage beyond the allowance is billed at the per-token rates below. On the plans page, Pro costs $10 per month, Pro+ $39 and Max $100, and Max includes $100 a month in AI Credits.

Premium request multipliers still exist only for Pro and Pro+ subscribers on an existing annual plan who stayed on request-based billing after June 1, 2026. For those subscribers, GitHub lists these multipliers for models still offered: Claude Haiku 4.5 at 0.33, MAI-Code-1.1-Flash at 0.25, GPT-5.3-Codex at 6, Claude Sonnet 4.6 at 9 and Claude Opus 4.8 at 27. Copilot code review carries a multiplier of 13 on that legacy billing.

Models by plan

Copilot Free and Copilot Student users get models only through automatic model selection. For paid plans, GitHub's documentation lists the following. Prices are per 1 million tokens at the default tier.

ModelProviderProPro+, Max, Business, EnterpriseInput / output price
GPT-6.1 SolOpenAINoYes$2.00 / $10.00
GPT-6 AstraOpenAINoYes$10.00 / $50.00
GPT-6 SolOpenAINoYes$2.00 / $10.00
GPT-6 LunaOpenAIYesYes$0.10 / $0.50
GPT-5.6 SolOpenAINoYes$4.00 / $20.00
GPT-5.6 TerraOpenAIYesYes$2.00 / $12.00
GPT-5.6 LunaOpenAIYesYes$0.20 / $1.20
GPT-5.3-CodexOpenAIYesYes$1.75 / $14.00
GPT-5.4 nanoOpenAINoPro+ and Max only$0.20 / $1.25
Claude Fable 5.1AnthropicNoYes$10.00 / $50.00
Claude Fable 5AnthropicNoYes$10.00 / $50.00
Claude Opus 5.5AnthropicNoYes$4.00 / $20.00
Claude Opus 5AnthropicNoYes$5.00 / $25.00
Claude Opus 4.8AnthropicNoYes$5.00 / $25.00
Claude Opus 4.8 (fast mode, preview)AnthropicNoYes$10.00 / $50.00
Claude Sonnet 5.5AnthropicYesYes$2.00 / $10.00
Claude Sonnet 5AnthropicYesYes$2.00 / $10.00
Claude Sonnet 4.6AnthropicAnnual plans onlyPro+ annual only$3.00 / $15.00
Claude Haiku 5.5AnthropicYesYes$0.10 / $0.50
Claude Haiku 4.5AnthropicYesYes$1.00 / $5.00
Gemini 4 ArgonGoogleNoYes$2.00 / $10.00
Gemini 3.8 FlashGoogleYesYes$0.75 / $3.75
Grok 4.7xAIYesYes$2.00 / $6.00
Grok 4.6xAIYesYes$2.00 / $6.00
MAI-Code-1.1-FlashMicrosoftYesYes$0.20 / $1.20
Kimi K3Moonshot AIYesYes$3.00 / $15.00

Several OpenAI, Google, Anthropic and xAI models cost more past a long-context threshold, for example above 272K input tokens for GPT-6.1 Sol. Pricing for Gemini 3.8 Flash and Gemini 4 Argon is promotional; GitHub says the Gemini 4 Argon promotion ends on December 1, 2026. Organization and enterprise owners can enable or restrict individual models for their members.

Recently retired models

ModelRetiredSuggested replacement
GPT-5.5October 19, 2026GPT-6.1 Sol
GPT-5.4October 19, 2026GPT-6.1 Sol
GPT-5.4 miniOctober 19, 2026GPT-6 Luna
GPT-5 miniOctober 19, 2026GPT-6 Luna
Gemini 3.7 FlashOctober 19, 2026Gemini 3.8 Flash
Grok 4.5October 19, 2026Grok 4.7
Claude Opus 4.7October 2, 2026Claude Opus 5.5
Gemini 3.5 FlashOctober 2, 2026Gemini 3.8 Flash
Gemini 3.6 FlashOctober 2, 2026Gemini 3.8 Flash
Kimi K2.7 CodeOctober 2, 2026Kimi K3
MAI-Code-1-FlashSeptember 10, 2026Latest MAI-Code model

Earlier, on September 1, 2026, GitHub retired Claude Opus 4.5, Claude Opus 4.6, Claude Sonnet 4.5, Claude Sonnet 4.6 (except for annual Pro and Pro+ subscribers), Gemini 3.1 Pro and Raptor mini.

Next retirements: November 16

GitHub plans to retire three more models on November 16, 2026: Claude Opus 4.8 and Claude Opus 5, both replaced by Claude Opus 5.5, and GPT-5.6 Terra, replaced by GPT-6.1 Sol or GPT-6 Luna. Claude Opus 4.8 will stay available to individual subscribers on annual plans. For Copilot Business and Copilot Enterprise, the suggested replacements are enabled automatically unless an administrator has turned off default model enablement.