OpenManifest
Connect Copilot
Catalog

37 models, one API key

Every model GitHub Copilot exposes through its chat endpoint, ordered by release date. The subset available to you depends on your Copilot subscription tier.

● Google · 4● Anthropic · 8● OpenAI · 23● xAI · 2
ModelInputOutput
gemini-3.5-flash
Gemini 3.5 Flash
Coding quality close to Gemini Pro at Flash-tier speed and cost. Strong tool use, high cache efficiency, 14× premium multiplier on Copilot. Pro / Pro+ / Business / Enterprise.
1M32K
claude-opus-4.7
Claude Opus 4.7
Anthropic's newest Opus. Long-horizon reasoning, multi-step tool execution, replaces 4.5/4.6 on Pro+. 15× premium multiplier (post-promo).
192K128K
gpt-5.5
GPT-5.5
OpenAI's newest flagship. Strongest reasoning + tool reliability in the GPT-5 line. Premium 7.5× multiplier on Copilot. Responses-API only.
400K128K
claude-opus-4.6
Claude Opus 4.6
Prior flagship Opus. Being phased out on Pro+ in favour of 4.7, still listed for Business/Enterprise and the Claude coding agent.
200K64K
claude-opus-4.6-fast
Claude Opus 4.6 (fast mode)
Preview fast-mode variant of Opus 4.6. Lower latency at the cost of some reasoning depth. Preview only.
200K64K
claude-sonnet-4.6
Claude Sonnet 4.6
Daily-driver Claude. Pairs with Opus 4.6 for cost-balanced agent fleets — excellent edit precision.
200K64K
raptor-mini
Raptor mini
xAI's compact Raptor model. Preview-only in Copilot. Built for high-throughput coding with very low latency.
256K32K
gemini-3.1-pro-preview
Gemini 3.1 Pro (Preview)
2M-context flagship Gemini. Preview only on Copilot. Native multimodal, structured-output strong.
2M64K
gpt-5.4
GPT-5.4
Step over 5.3-Codex with stronger general reasoning. Solid daily driver for whole-repo refactors.
400K128K
gpt-5.4-mini
GPT-5.4 mini
Mid-tier 5.4 — faster than 5.4 base with similar context. Responses-API only on Copilot.
400K64K
gpt-5.4-nano
GPT-5.4 nano
Compact 5.4 sibling. Copilot Pro+ exclusive, currently surfaced only in the Codex VS Code extension.
200K64K
gemini-3-flash-preview
Gemini 3 Flash (Preview)
Fast Gemini 3 tier. 1M context with low latency. Preview only on Copilot.
1M32K
gpt-5.3-codex
GPT-5.3 Codex
Code-tuned 5.3. Specialised for diff-shaped output and multi-file edits. Responses-API only.
400K128K
claude-sonnet-4.5
Claude Sonnet 4.5
Workhorse Sonnet. Mature tool protocol; the most predictable Claude for code-edit benchmarks.
200K64K
claude-opus-4.5
Claude Opus 4.5
Prior-gen Opus. Still strong for deep refactors when latency or cost matter more than raw frontier capability.
200K64K
claude-haiku-4.5
Claude Haiku 4.5
Smallest, fastest Claude 4. Cheap completions, autocomplete loops, and high-volume agent fan-out.
200K32K
gpt-5.2
GPT-5.2
Mid-cycle 5.x. Balances cost and reasoning, popular in agent loops.
400K128K
gpt-5.2-codex
GPT-5.2 Codex
5.2 fine-tuned for code edits. Strong at staying in-protocol over long agent sessions. Responses-API only.
400K128K
grok-code-fast-1
Grok Code Fast 1
Code-tuned Grok. 0× premium multiplier on Copilot — effectively free for high-volume experimentation. Zero data-retention upstream.
256K32K
gpt-5.1
GPT-5.1
First incremental over GPT-5. Improved tool-call reliability and lower latency.
400K128K
gpt-5.1-codex
GPT-5.1 Codex
5.1 Codex variant. Stronger refactor and bug-fix benchmarks than 5.1 base.
400K128K
gpt-5.1-codex-max
GPT-5.1 Codex Max
Heavyweight 5.1 Codex. Built for very long agentic sessions and architectural review.
400K128K
gpt-5.1-codex-mini
GPT-5.1 Codex Mini
Compact 5.1 Codex. High throughput, low latency for autocomplete-style loops.
200K64K
gpt-5
GPT-5
Initial GPT-5 release. Long context, tool-aware, strong code synthesis. Baseline reference for the 5.x series.
400K128K
gpt-5-mini
GPT-5 mini
Lighter, faster GPT-5 sibling. Excellent default for high-throughput coding agents.
400K128K
claude-sonnet-4
Claude Sonnet 4
First-gen Claude 4 Sonnet. Stable tool-use protocol, broadly tested across coding agents.
200K64K
gpt-4.1
GPT-4.1
1M-context successor to GPT-4o. Best in class for whole-repo grounding and long-form refactors.
1M32K
gpt-4.1-2025-04-14
GPT-4.1 (2025-04-14 snapshot)
Pin GPT-4.1 to its launch-day weights. Use when you need bit-identical replay of older runs.
1M32K
gemini-2.5-pro
Gemini 2.5 Pro
Prior flagship Gemini. 1M context, mature tool-use, dependable for structured output.
1M64K
gpt-4o-2024-11-20
GPT-4o (2024-11-20 snapshot)
Most recent GPT-4o snapshot. Pin when you need stable output across deploys.
128K16.384K
gpt-4o-2024-08-06
GPT-4o (2024-08-06 snapshot)
Mid-2024 GPT-4o snapshot. Introduced structured outputs / strict JSON mode.
128K16.384K
gpt-4o-mini
GPT-4o mini
Lighter GPT-4o for high-volume completions. Same 128K context, much cheaper per token.
128K16.384K
gpt-4o-mini-2024-07-18
GPT-4o mini (2024-07-18 snapshot)
GPT-4o mini snapshot. Cheaper batch jobs and high-throughput agents.
128K16.384K
gpt-4o
GPT-4o
OpenAI's omni-model. Fast, multimodal, broadly supported across legacy coding tools.
128K16.384K
gpt-4o-2024-05-13
GPT-4o (2024-05-13 launch)
Original GPT-4o launch weights. Lower output cap (4K) and no structured outputs.
128K4.096K
gpt-4
GPT-4
Original GPT-4. Kept for backwards-compatibility with older agents pinned to this id.
8.192K4.096K
gpt-3.5-turbo
GPT-3.5 Turbo
Legacy chat model. Cheap and fast; useful for trivial routing/classification work.
16.385K4.096K