Public benchmarkExternal source reviewReviewed by DanPublished Source checked Updated
Astra 6.2: new model and public task-evaluation evidence
The report
Verdict
Published benchmark evidence
Cognition lists Astra 6.2 at 44.2% and $5.97 per rollout. Sol 6.2 is listed at 43.8% and $1.29: a similar published score with substantially different estimated API spend.
On the tier list: The November 17 tier is Dan’s own ranking; public benchmark numbers are a separate evidence channel, not an automatic tier adjustment.
What the benchmarks report
- FrontierCode 2.0 — Cognition: 44.2%. Astra 6.2; max; Codex harness. Reported by Cognition benchmark operator.
- FrontierCode 2.0 mean rollout cost: $5.97. Astra 6.2; max; benchmark API-price estimate. Reported by Cognition benchmark operator.
- Terminal-Bench-Science 0.1 vs Astra 6: 9 percentage points higher. Astra 6.2 relative to GPT-6 Astra; publisher comparison. Reported by OpenAI vendor release.
Original sources
Source 1 — FrontierCode 2.0 — Cognition, FrontierCode 2.0 mean rollout cost.
Source 2 — Terminal-Bench-Science 0.1 vs Astra 6.
What this evidence tells a reader
These Cognition entries allow a narrower comparison than most mixed-benchmark charts: both name FrontierCode 2.0, max effort and a Codex harness. Astra's 44.2% versus Sol's 43.8% is a 0.4-percentage-point difference in the displayed score. Without repeated-run variation, that number is not evidence of a meaningful reliability advantage.
Cognition describes cost as mean USD spend per rollout at list API rates. At $5.97 versus $1.29, Astra's published figure is about 4.6 times Sol's, yet this is not a quote for DanManREAL's subscription, contract, or one actual coding job. The cost/score tradeoff deserves evaluation when tasks have a strict budget.
OpenAI's reported nine-percentage-point science-terminal improvement is a comparison to an earlier Astra version on a separate scientific evaluation. It should not be treated as nine points above Sol on FrontierCode, or as an independently reproduced result.
How to use these results
- If both models are accessible in the same workflow, start with a representative patch and compare actual cost, diff quality and review effort.
- Use the Cognition pair for a provisional API-cost discussion, not for predicting subscription charges.
- Treat scientific-terminal claims as a separate evidence track from repository mergeability.
Limits of this report
- The published 0.4-point score gap has no reported confidence interval or repeatability data here.
- A $5.97 benchmark rollout price and vendor science comparison are not measured workshop spend or a guarantee of production speed.
Follow the lab
New test reports and tier-list changes get announced in my Discord's #news channel. The full videos are on YouTube.
Receipts
DanManREAL · AI lab
Test receipt
GPT-6.2 Astra
- FrontierCode 2.0 — Cognition
- 44.2%Cognition benchmark operator (independent operator or secondary publication). Astra 6.2; max; Codex harness. Not independently rerun by DanManREAL.
- FrontierCode 2.0 mean rollout cost
- $5.97Cognition benchmark operator (independent operator or secondary publication). Astra 6.2; max; benchmark API-price estimate. Not independently rerun by DanManREAL.
- Terminal-Bench-Science 0.1 vs Astra 6
- 9 percentage points higherOpenAI vendor release (model publisher). Astra 6.2 relative to GPT-6 Astra; publisher comparison. Not independently rerun by DanManREAL.
- Source review
- November 17, 2026Date these source claims were reviewed; not a claim about when the published benchmark was executed.
Verdict
Published benchmark evidence
Tier list The November 17 tier is Dan’s own ranking; public benchmark numbers are a separate evidence channel, not an automatic tier adjustment.
danmanreal.com/tests/gpt-6-2-astra-public-evidence
Models covered by external benchmarks
On the tier list's bench
- GPT-6.2 Astra · 1 receipt on file
Related
Independent testing. Sponsorships are always disclosed, and verdicts are never for sale.
Disagree? Bring your receiptsAll reports
- One inbox for my AI coding agents: triage took 34% less time, resuming work did not improveHead-to-headNov 20
- Mythos 5.2: research and security benchmark source reviewPublic benchmarkNov 17
- Fable 5.2: agentic coding benchmark source reviewPublic benchmarkNov 17
- Claude Opus 5.5: public coding benchmark receiptsPublic benchmarkNov 17
- Astra 6.2: new model and public task-evaluation evidencePublic benchmarkNov 17
- Sol 6.2: new model and public coding benchmark evidencePublic benchmarkNov 17
- Gemini 4 Carbon: benchmark source and usage profilePublic benchmarkNov 17
- Qwen 4: Qwen4-Max launch and benchmark sourcesPublic benchmarkNov 17
- Kimi K3.1 Pro: public coding benchmark receiptsPublic benchmarkNov 17
- GLM-5.3: public coding benchmark receiptsPublic benchmarkNov 17
- Gemini 4 Argon: public coding benchmark receiptsPublic benchmarkNov 17
- Claude Sonnet 5.5: public coding benchmark receiptsPublic benchmarkNov 17
- DeepSeek V4.1 Flash: public coding benchmark receiptsPublic benchmarkNov 17
- Muse Wonder Full 1.0: public coding benchmark receiptsPublic benchmarkNov 17
- Grok 4.7: separate release and public coding resultsPublic benchmarkNov 17
- ai-design-check on two of my FABLE screens: 10 warnings, 4 of them false alarmsField testNov 2
- Vercel Agent Desktop used a sample Trip app in 6 actions without moving the mouseField testNov 1
- Cerberus Guard on sample data: one crawler name, two verdicts, and a block earned by behaviorField testOct 31
- GPT-6.1 Terra vs Opus 5.5: three real projects, and the designs I picked blindHead-to-headOct 29
- Steps → Reduced: opening a terminal agent went from 98 seconds to 11AuditOct 28
- One ChatGPT chat directed Opus and Codex at once through DMR_REMOTEField testOct 28
- Claude Duet ran my channel for a week and caught 269 fake AI crawlersField testOct 27
- Kimi K3.1 Pro built my animation family, after a quota stop and 21 minutes of failed requestsField testOct 24
- ChatGPT + Codex built 33 new DMR_REMOTE toolsField testOct 23
- Gemini 4 Argon reviewed a finished episode: 14 clips, one flag, one inconsistencyField testOct 21
- Qwen 3.8 Flash Next on my RTX 5080: 81 tokens a second with 202K tokens readAuditOct 21
- Does an extra AI review step pay off? A small DeepSeek trial said noHead-to-headOct 19
- My real DeepSeek bill: 12.31 billion tokens for $122.28Cost receiptOct 17
- Gemini 4 Argon kept hanging in Gemini CLI. The cause was a 60-second timeoutIncidentOct 14
- I asked 10 AI coding agents for their biggest weaknessSelf-reportOct 14
- Jarvis 27B on my own PC: 30 tokens a second, and it still recalls at 251KAuditOct 10
- My agents' first game benchmark was wrong. Here's how it was caughtIncidentOct 10
- Pixal3D + Blender turned my 2D art into animated 3D charactersField testSep 22
- DeepSeek Flash answered 200 OK and sent nothing for two hoursIncidentSep 14
- Grok, Codex and DeepSeek played WorldOS for Rank 1. Grok won with 98Head-to-headAug 20
- DeepSeek V4 Flash audited 18 of my real coding sessions. Two said 'done' too earlyAuditAug 4