Compare / head-to-head
DeepSeek V3.2-Expvs
GPT-6 Astra
GPT-6 Astra leads 4 of 4 shared benchmarks. GPT-6 Astra has the larger context window (1.1M tokens).
Benchmarks from cited public sources; pricing from official pages; status from official provider feeds.
Shared benchmarks
0 – 4
GPT-6 Astra leads
Cheaper per token
—
list price, input + output
Larger context
GPT-6 Astra
1.1M tokens
Providers
2 providers
DeepSeek · OpenAI
DeepSeek V3.2-Exp
DeepSeek · released 2025-09-29
text
- Context
- 128K
- Max out
- —
- Input /1M
- —
- Output /1M
- —
- Cached /1M
- —
- Scores
- 17 · 7 core
GPT-6 Astra
OpenAI · released 2026-09-03
textvision
Quality
Benchmark matrix
4 shared · 13 only DeepSeek V3.2-Exp · 29 only GPT-6 Astra| Benchmark | DeepSeek V3.2-Exp | GPT-6 Astra | Δ | Edge |
|---|---|---|---|---|
| Reported by both · 4 | ||||
| LMArena Elo | 1424.9 ↗ | 1477.1 ↗ | -52.2 | GPT-6 Astra |
| LMArena WebDev (LMArena) | 1271.8 ↗ | 1787.7 ↗ | -515.9 | GPT-6 Astra |
| Vending-Bench 2 (Andon Labs) | 1034 ↗ | 15514.7 ↗ | -14480.7 | GPT-6 Astra |
| WeirdML (Håvard Tveit Ihle) | 39.5% ↗ | 93.6% ↗ | -54.1 pt | GPT-6 Astra |
| Only DeepSeek V3.2-Exp reports · 13 | ||||
| MMLU-Pro | 85% ↗ | not reported | — | — |
| AIME 2025 | 89.3% ↗ | not reported | — | — |
| Humanity's Last Exam (no tools) | 19.8% ↗ | not reported | — | — |
| Aider-Polyglot | 74.5% ↗ | not reported | — | — |
| BrowseComp | 40.1% ↗ | not reported | — | — |
| BrowseComp-zh | 47.9% ↗ | not reported | — | — |
| GPQA-Diamond | 79.9% ↗ | not reported | — | — |
| HMMT 2025 | 83.6% ↗ | not reported | — | — |
| LiveCodeBench | 74.1% ↗ | not reported | — | — |
| SimpleQA (agentic tool use) | 97.1% ↗ | not reported | — | — |
| SWE Verified | 67.8% ↗ | not reported | — | — |
| SWE-bench Multilingual | 57.9% ↗ | not reported | — | — |
| Terminal-bench | 37.7% ↗ | not reported | — | — |
| Only GPT-6 Astra reports · 29 | ||||
| GPQA Diamond | not reported | 96% ↗ | — | — |
| APEX-Agents (Mercor) | not reported | 64.7% ↗ | — | — |
| AutomationBench 1.0.6 | not reported | 41.4% ↗ | — | — |
| BALROG (BALROG) | not reported | 68.3% ↗ | — | — |
| Chess Puzzles (Epoch AI run) | not reported | 72% ↗epoch run | — | — |
| DeepSWE v1.1 (Datacurve) | not reported | 74.1% ↗ | — | — |
| EBR-bench (Epoch AI run) | not reported | 76.2% ↗epoch run | — | — |
| FrontierMath Tier 4 v2 (Epoch AI run) | not reported | 97.6% ↗epoch run | — | — |
| FrontierMath Tiers 1-3 v2 (Epoch AI run) | not reported | 93.7% ↗epoch run | — | — |
| FrontierSWE V2 (Proximal Labs) | not reported | 65.5% ↗ | — | — |
| Furniture Assembly (Epoch AI run) | not reported | 80% ↗epoch run | — | — |
| GSO Opt@1 (GSO) | not reported | 77.5% ↗ | — | — |
| LiveBench Agentic Coding (LiveBench) | not reported | 57.3% ↗ | — | — |
| LiveBench Coding (LiveBench) | not reported | 80.4% ↗ | — | — |
| LiveBench Data Analysis (LiveBench) | not reported | 83% ↗ | — | — |
| LiveBench Instruction Following (LiveBench) | not reported | 75.6% ↗ | — | — |
| LiveBench Language (LiveBench) | not reported | 89.4% ↗ | — | — |
| LiveBench Mathematics (LiveBench) | not reported | 96.8% ↗ | — | — |
| LiveBench Reasoning (LiveBench) | not reported | 92.7% ↗ | — | — |
| LMArena Agent (LMArena) | not reported | 0.1227 ↗ | — | — |
| LMArena Vision (LMArena) | not reported | 1284.3 ↗ | — | — |
| MirrorCode (Epoch AI run) | not reported | 46.7% ↗epoch run | — | — |
| Mystery Game Puzzles (Epoch AI run) | not reported | 84% ↗epoch run | — | — |
| OTIS Mock AIME 2024-2025 (Epoch AI run) | not reported | 100% ↗epoch run | — | — |
| SAGE (Vals AI) | not reported | 46.4% ↗ | — | — |
| SimpleBench (SimpleBench) | not reported | 86.5% ↗ | — | — |
| SimpleQA Verified | not reported | 75.6% ↗epoch run | — | — |
| Terminal-Bench 4.0 (Vals AI) | not reported | 59.6% ↗ | — | — |
| Terminal-Bench Science 0.1 | not reported | 64.6% ↗ | — | — |
Scores tagged "epoch" or "matharena" are independent runs, used only where the lab has not published its own; ⚠ marks rows MathArena flags as released after the competition. Higher is better on every row. Δ is DeepSeek V3.2-Exp minus GPT-6 Astra in the benchmark's own unit. "Not reported" means the lab has not published that figure; it is not a zero. ↗ opens the source.
Specs & pricing
Side by side
Official pricing pages and model cards| Spec | DeepSeek V3.2-Exp | GPT-6 Astra | Edge |
|---|---|---|---|
| Context window | 128K tokens | 1.1M tokens | GPT-6 Astra |
| Max output | — | 128K tokens | — |
| Input price / 1M | — | $10 ↗ | — |
| Output price / 1M | — | $50 ↗ | — |
| Cached input / 1M | — | $1 ↗ | — |
| Input + output / 1M Lower is cheaper. List prices; batch, tool and regional fees excluded. | — | $60 | — |
| Modalities | text | text · vision | GPT-6 Astra |
| Released | 2025-09-29 | 2026-09-03 | — |
| Cited benchmark scores | 17 | 39 | — |
Reliability
Provider status
Live from /statusMore matchups
DeepSeek V3.2-Exp vs …
Models sharing the most benchmarksMore matchups
GPT-6 Astra vs …
Models sharing the most benchmarksBuilt by Respan
Which one wins on your data?
Public benchmarks are a starting point. Run DeepSeek V3.2-Exp and GPT-6 Astra on your own prompts with Respan evals, or route to either through one gateway key with automatic failover.