GPT-6 Astra vs Claude Sonnet 5
Choose two or three models. Copy the URL to share your selection.
| Comparison | GPT-6 Astragpt-6-astra | Claude Sonnet 5claude-sonnet-5 |
|---|---|---|
| Type | Chat | Chat |
| Best for | Chat, writing, coding and reasoning tasks | Chat, writing, coding and reasoning tasks |
| TokenRex price | $10.00 in / $50.00 out per 1M tokens | $2.00 in / $10.00 out per 1M tokens |
| Cached input / 1M tokens | $1.00 per 1M tokens | $0.20 per 1M tokens |
| Cache write / 1M tokens | $12.50 per 1M tokens | $2.50 per 1M tokens |
| Cached output | Not offered · standard output rate applies | Not offered · standard output rate applies |
| Market price | $10.00 in / $50.00 out per 1M tokens | $2.00 in / $10.00 out per 1M tokens |
| Context window (provider) | 1,050,000 tokens | 1,000,000 tokens |
| Input token limit (provider) | Within the context window | Within the context window |
| Maximum output | 128,000 tokens | 128,000 tokens |
| Knowledge cutoff | Apr 30, 2026 | Jan 2026 |
| Vendor input modalities | Text, Image | Text, Image |
| Vendor output modalities | Text | Text |
| TokenRex endpoints | POST /v1/chat/completions | POST /v1/chat/completions · POST /v1/messages |
| TokenRex input | Text, Image/PDF content through the chat bridge | Text, Image/PDF content through the chat bridge |
| TokenRex output | Text | Text |
| Benchmarks | ||
| Benchmark evidence | 6 source-linked results | 4 source-linked results |
| FrontierMath Tier 4 (v2) | 97.6% Third-party reproduction of OpenAI’s announcement table; reported maximum effort unless noted, not a TokenRex run. Direct launch page could not be fetched during this audit. Vellum · GPT-6 Astra launch-table analysis | — |
| ARC-AGI-3 | 99.9% Semi-private ARC-AGI-3, Provider Adapter harness, high reasoning: 99.9%. Standard harness at max reasoning instead reports 62.7%; opaque state reuse and compaction materially change the result. ARC Prize · GPT-6 Astra evaluated results | — |
| ExploitBench | 100% Third-party reproduction of OpenAI’s announcement table; reported maximum effort unless noted, not a TokenRex run. Direct launch page could not be fetched during this audit. Evaluated without production safeguards. Vellum · GPT-6 Astra launch-table analysis | — |
| Terminal-Bench-Science 0.1 | 64.6% Third-party reproduction of OpenAI’s announcement table; reported maximum effort unless noted, not a TokenRex run. Direct launch page could not be fetched during this audit. Vellum · GPT-6 Astra launch-table analysis | — |
| FrontierCode v1.1 (Extended) | 60.25% Cognition FrontierCode 1.1; GPT-6 Astra; medium effort; codex harness; Extended subset (150 tasks). Published weighted rubric score (new_score × 100), not the all-or-nothing pass rate; solution-source violations score zero. Source field new_score=0.6025. Publisher evaluation, not a TokenRex measurement; provider effort configuration is not a guarantee of the API's fixed thinking budget. Cognition · FrontierCode 1.1 original leaderboard | 49.28% Cognition FrontierCode 1.1; Claude Sonnet 5; medium effort; claude-code harness; Extended subset (150 tasks). Published weighted rubric score (new_score × 100), not the all-or-nothing pass rate; solution-source violations score zero. Source field new_score=0.4928. Publisher evaluation, not a TokenRex measurement; provider effort configuration is not a guarantee of the API's fixed thinking budget. Cognition · FrontierCode 1.1 original leaderboard |
| FrontierCode v1.1 (Main) | 48.83% Cognition FrontierCode 1.1; GPT-6 Astra; medium effort; codex harness; Main subset (100 tasks). Published weighted rubric score (new_score × 100), not the all-or-nothing pass rate; solution-source violations score zero. Source field new_score=0.4883. Publisher evaluation, not a TokenRex measurement; provider effort configuration is not a guarantee of the API's fixed thinking budget. Cognition · FrontierCode 1.1 original leaderboard | 35.2% Cognition FrontierCode 1.1; Claude Sonnet 5; medium effort; claude-code harness; Main subset (100 tasks). Published weighted rubric score (new_score × 100), not the all-or-nothing pass rate; solution-source violations score zero. Source field new_score=0.352. Publisher evaluation, not a TokenRex measurement; provider effort configuration is not a guarantee of the API's fixed thinking budget. Cognition · FrontierCode 1.1 original leaderboard |
| GMMLU | — | 89% Fable 5.1 system card section 8.18 comparison table: adaptive thinking max effort, single trial, no tools or custom system prompts. Published comparison, not catalog medium effort. Anthropic Fable 5.1 system card · sections 8.1–8.11 |
| MILU | — | 89.3% Fable 5.1 system card section 8.18 comparison table: adaptive thinking max effort, single trial, no tools or custom system prompts. Published comparison, not catalog medium effort. Anthropic Fable 5.1 system card · sections 8.1–8.11 |
No overall score, winner or cost-versus-score frontier is calculated without comparable, attributable benchmark evidence. Different test configurations appear separately. Vendor capabilities are distinct from TokenRex API support.
— means no published value or no matching capability. Provider token limits are not tested gateway guarantees; TokenRex media controls and response formats are listed separately.
GPT-6 Astra
Sources and verification
Provider specifications describe the upstream model, not a guarantee of every feature through TokenRex. Prices come from the enabled catalog; benchmark results belong to the exact tested model.
- Provider model specifications
Vendor-reported
Last checked - Vellum · GPT-6 Astra launch-table analysis
Third-party report
Last checked - ARC Prize · GPT-6 Astra evaluated results
Third-party report
Last checked - Cognition · FrontierCode 1.1 original leaderboard
Third-party report
Last checked
Research notes (1)
- Published provider limits; maximum-capacity requests have not been independently exercised through TokenRex. Thinking-level variants use the same provider model; reasoning and answer tokens share the output budget.
Claude Sonnet 5
Sources and verification
Provider specifications describe the upstream model, not a guarantee of every feature through TokenRex. Prices come from the enabled catalog; benchmark results belong to the exact tested model.
- Provider model specifications
Vendor-reported
Last checked - Anthropic Fable 5.1 system card · sections 8.1–8.11
Vendor-reported
Last checked - Cognition · FrontierCode 1.1 original leaderboard
Third-party report
Last checked
Research notes (1)
- Published provider limits; maximum-capacity requests have not been independently exercised through TokenRex. Thinking-level variants use the same provider model; reasoning and answer tokens share the output budget.
Popular comparisons
- GPT-6 Sol vs Claude Opus 5.5
- GPT-6 Sol vs Gemini 4 Argon
- Claude Opus 5.5 vs Gemini 4 Argon
- GPT-6.1 Sol vs GPT-6 Sol
- GPT-6 Astra vs Claude Sonnet 5
- GPT-6 Astra vs Gemini 3.1 Pro
- Claude Sonnet 5 vs Gemini 3.1 Pro
- Claude Opus 5.5 vs Claude Sonnet 5
- GPT-5.5 vs Claude Sonnet 5
- GPT-5.5 vs Gemini 2.5 Pro
- GPT-5 vs Gemini 2.5 Pro
- GPT-6 Luna vs Gemini 3.8 Flash
- Claude Haiku 4.5 vs Gemini 3.8 Flash
- GPT-5 Mini vs Gemini 2.5 Flash
- GPT-5 Nano vs Gemini 2.5 Flash Lite
- GPT-5.4 Mini vs Claude Haiku 4.5