Model comparison
GLM 5.3 vs GPT-6 Astra
Z.ai against OpenAI, compared on context, price, and attributed benchmark results.
Recorded profile verification October 2, 2026Individual provider fields may changeEvidence confidence: Medium — see receipts below
Z.ai
GLM 5.3
Frontier · Open weights
OpenAI
GPT-6 Astra
Frontier
| Specification | GLM 5.3 | GPT-6 Astra |
|---|---|---|
| Provider | ||
| Provider | Z.ai | OpenAI |
| Tier | ||
| Tier | Frontier | Frontier |
| Context window | ||
| Context window | 1M | Winner: 1.05M |
| Max output | ||
| Max output | Winner: 131K | 128K |
| Input / 1M tokens | ||
| Input / 1M tokens | Winner: $1.40 | $10 |
| Output / 1M tokens | ||
| Output / 1M tokens | Winner: $4.40 | $50 |
| Weights | ||
| Weights | Open | Closed |
| Parameters | ||
| Parameters | 753B total (MoE; active count not verified) | Not verifiedUnverified |
| Reasoning levels | ||
| Reasoning levels | low, high, max | low, medium, high, xhigh, max |
| Modalities | ||
| Modalities | text | text, image |
| License | ||
| License | GLM-5.3 License (custom) | Not verifiedUnverified |
| API model id | ||
| API model id | glm-5.3 | gpt-6-astra |
| Released | ||
| Released | August 14, 2026 | September 3, 2026 |
| Artificial Analysis Intelligence Index [max] · source date GLM 5.3: 2026-10-01 | ||
| Artificial Analysis Intelligence Index [max] · source date GLM 5.3: 2026-10-01 | 45 | Not verifiedUnverified |
| Artificial Analysis Intelligence Index [max] · source date GPT-6 Astra: 2026-10-01 | ||
| Artificial Analysis Intelligence Index [max] · source date GPT-6 Astra: 2026-10-01 | Not verifiedUnverified | 53 |
| Terminal-Bench 2.1 · source date GLM 5.3: 2026-08-14 | ||
| Terminal-Bench 2.1 · source date GLM 5.3: 2026-08-14 | 88.2 | Not verifiedUnverified |
| DeepSWE 1.1 · source date GLM 5.3: 2026-08-14 | ||
| DeepSWE 1.1 · source date GLM 5.3: 2026-08-14 | 66.9 | Not verifiedUnverified |
| DeepSWE 1.1 · source date GPT-6 Astra: 2026-09-03 | ||
| DeepSWE 1.1 · source date GPT-6 Astra: 2026-09-03 | Not verifiedUnverified | 74.1 |
| Humanity's Last Exam · source date GLM 5.3: 2026-08-14, GPT-6 Astra: 2026-09-03 (comparability not verified) | ||
| Humanity's Last Exam · source date GLM 5.3: 2026-08-14, GPT-6 Astra: 2026-09-03 (comparability not verified) | 62.5 | 57.2 |
| OSWorld 2.0 (offline) · source date GPT-6 Astra: 2026-09-03 | ||
| OSWorld 2.0 (offline) · source date GPT-6 Astra: 2026-09-03 | Not verifiedUnverified | 72.6 |
| GPQA Diamond · source date GPT-6 Astra: 2026-09-03 | ||
| GPQA Diamond · source date GPT-6 Astra: 2026-09-03 | Not verifiedUnverified | 96 |
Dates beside benchmark scores refer to the linked source’s publication or capture date unless that source explicitly states when it measured the result. A shared URL alone does not establish comparable runs; unverified setups are marked. Prices are USD per million tokens at standard rates, excluding batch and caching discounts. Values we could not confirm from the provider are shown as “Not verified” rather than estimated.
Benchmark receipts
Where each score comes from, and how far it can be compared across models.
- Source date2026-10-01Artificial Analysis (GPT-6 Astra (max))
Composite index blending reasoning, knowledge, and coding evals into one 0–100 score.
Comparable with caveatOnly same-version scores are comparable. v4.3.2 was rebaselined (the top score fell from 66 on v4.1.1 to ~58), so v4.1.1 figures must not be compared with v4.3.2 figures — check the cited source and its date.
- Source date2026-10-01Artificial Analysis (GLM-5.3 (max))
Composite index blending reasoning, knowledge, and coding evals into one 0–100 score.
Comparable with caveatOnly same-version scores are comparable. v4.3.2 was rebaselined (the top score fell from 66 on v4.1.1 to ~58), so v4.1.1 figures must not be compared with v4.3.2 figures — check the cited source and its date.
- Source date2026-09-03OpenAI GPT-6 Astra evaluation table (vendor-reported, with tools; maximum across evaluated efforts)
Frontier-knowledge ceiling: extremely hard multi-domain questions written to be near-impossible without deep expertise.
Comparable with caveatSubset/tool configurations (text-only vs with-tools) differ between vendors and materially change scores; must match configuration to compare.
- Source date2026-09-03OpenAI GPT-6 Astra evaluation table (vendor-reported; maximum across evaluated efforts)
Graduate-level science reasoning: multiple-choice questions written by domain PhDs.
Comparable with caveatDo not treat a fixed two-point band as universal measurement uncertainty; it depends on run count and evaluation setup. Chain-of-thought, prompt, and tool policy must match. The small pool limits fine-grained conclusions, and frontier scores can be near ceiling; Artificial Analysis dropped it from its Intelligence Index in v4.3.
- Source date2026-09-03OpenAI (vendor-reported) (v2026.08.08 offline set, partial reward; vendor research/API evaluation, maximum across evaluated efforts)
Computer-use: completing real OS tasks (files, apps, settings) in a sandboxed desktop.
Comparable with caveatOffline vs live interaction modes are NOT comparable; step budgets and screen resolution must match.
- Source date2026-09-03OpenAI (xhigh, mini-swe-agent; matches Datacurve's independent board (74.1% at $4.43/task, rank 1; board updated 2026-09-22))
Agentic software engineering: pass@1 on real repo tasks with reported cost per completed task.
Directly comparable
- Source date2026-08-14Z.ai GLM-5.3 blog + HF model card (vendor, with tools, full set)
Frontier-knowledge ceiling: extremely hard multi-domain questions written to be near-impossible without deep expertise.
Comparable with caveatSubset/tool configurations (text-only vs with-tools) differ between vendors and materially change scores; must match configuration to compare.
- Source date2026-08-14Z.ai GLM-5.3 blog + HF model card (vendor, mini-swe-agent family)
Agentic software engineering: pass@1 on real repo tasks with reported cost per completed task.
Directly comparable
- Source date2026-08-14Z.ai GLM-5.3 blog + HF model card eval results (vendor, Claude Code harness)
Agentic terminal work: multi-step tasks executed in a sandboxed shell environment.
Comparable with caveatNot comparable with Terminal-Bench 3.0 or 4.0 (different task sets) or v1; harness configuration (container, time limits) also shifts results.
Pricing tiers
GLM 5.3: Z.ai token API $1.40/$4.40 per MTok; cached input $0.26. Cache storage is listed as limited-time free, with no verified end date. Coding Plan points are separate: off-peak/weekend calls consume half the standard points. API docs advertise 1M context and specify a 131,072 output ceiling.
GPT-6 Astra: Standard $10/$50 per MTok (≤272K input); above 272K the full request bills $20/$75. Cached input $1/$2 and cache writes $12.50/$25 for short/long context. Batch/Flex are half Standard; Fast is 2x. Ultrafast is 6x: $60/$300 short, $120/$450 long, through Responses with global/US processing; no EU or other regional inference residency. Regional/FedRAMP processing adds 10% where eligible. Knowledge cutoff Apr 30 2026. Enterprise model access is disabled by default.
GLM 5.3
Z.ai's text-only GLM flagship, with a 753B-total MoE checkpoint and always-enabled reasoning. Published benchmark figures identify particular harnesses and settings.
Best for
- Coding Plan subscribers
- Long-horizon coding
- Chinese + English
Watch out
Custom licence: model-as-a-service operators whose licensee/affiliate aggregate revenue exceeds $10B in any consecutive 12 months require Z.AI security review before commercial use; the definition excludes embedded end-user features/harnesses and mere relays. Keep attribution/permission notices. Reasoning defaults to max; disabling it fails. Accounts that ever subscribed to Coding Plan, including expired subscriptions, currently have Model API access only through the Chat Completions-compatible protocol. Active parameter count and exact initial weight-publication day remain unverified.
GPT-6 Astra
OpenAI describes GPT-6 Astra as its most capable model for complex reasoning, coding and computer use, and the first it gates at the Preparedness Framework's 'Critical' cybersecurity threshold.
Best for
- Computer-use agents
- Agentic coding
- Hard reasoning
Watch out
OpenAI places Astra at its Critical cybersecurity threshold and disables it by default for Enterprise. Tool calling requires Responses; none reasoning and custom temperature/top_p/logprobs are unsupported. Long prompts above 272K increase the full-request rate. GPT-6.1 Sol has lower base token prices; compare results on your own workload.
Benchmark
DeepSWE 1.1 in context
Only GPT-6 Astra has a published DeepSWE 1.1 result. It is shown against the wider field, with cost per completed task alongside the score.
Local leader
GPT-6 Astra [xhigh]
74.1%
Rows shown
28
Best source-matched Pass@1 per model
Snapshot date
2026-10-01
Mirrored from deepswe.datacurve.ai
Better is toward the top-right (higher pass rate, lower cost). X-axis is reversed to match DeepSWE’s public chart. v1.1 uses average cost, output tokens and steps. Legacy v1 transcriptions are excluded here until an exact primary receipt is retained.
| # | Model | Pass@1 | Cost / task | Tokens / task | Steps / task |
|---|---|---|---|---|---|
| 1 | GPT-6 Astra [xhigh] | 74.1% | $4.43 | 30k | 29 |
| 2 | Gemini 3.8 Flash [high] | 73.8% | $2.36 | 143k | 166 |
| 3 | Claude Opus 5 [max] | 73.6% | $11.84 | 118k | 99 |
| 4 | GPT-5.6 Sol [max] | 72.7% | $8.39 | 60k | 61 |
| 5 | Claude Fable 5 [xhigh] | 69.9% | $13.41 | 80k | 68 |
| 6 | GPT-5.6 Terra [max] | 69.6% | $4.95 | 72k | 76 |
| 7 | GLM 5.3 [max] | 69.0% | $3.99 | 80k | 124 |
| 8 | Kimi K3 [max] | 68.5% | $4.65 | 82k | 98 |
| 9 | Grok 4.6 [medium] | 67.5% | $3.45 | 50k | 70 |
| 10 | GPT-5.6 Luna [max] | 67.2% | $3.03 | 73k | 102 |
| 11 | GPT-5.5 [xhigh] | 67.0% | $7.23 | 46k | 82 |
| 12 | Gemini 3.7 Flash [medium] | 65.5% | $2.03 | 94k | 117 |
| 13 | GLM 5.3 Flash [max] | 63.4% | $0.48 | 73k | 123 |
| 14 | DeepSeek V4-Pro [max] | 62.8% | $0.24 | 106k | 155 |
| 15 | Claude Opus 4.8 [max] | 59.0% | $13.22 | 135k | 120 |
| 16 | Qwen3.8-Max [xhigh] | 57.5% | $3.73 | 95k | 111 |
| 17 | Muse Spark 1.2 [xhigh] | 54.9% | $3.70 | 99k | 101 |
| 18 | Claude Sonnet 5 [max] | 53.8% | $26.40 | 214k | 268 |
| 19 | Grok 4.5 [high] | 53.8% | $2.42 | 36k | 61 |
| 20 | DeepSeek V4-Flash [max] | 53.3% | $0.10 | 108k | 153 |
| 21 | Muse Spark 1.1 [xhigh] | 53.3% | $2.36 | 74k | 96 |
| 22 | GPT-5.4 [xhigh] | 51.8% | $5.65 | 71k | 70 |
| 23 | Gemini 3.6 Flash [high] | 46.7% | $4.42 | 96k | 117 |
| 24 | GLM 5.2 [max] | 43.8% | $3.92 | 78k | 129 |
| 25 | Gemini 3.5 Flash [high] | 36.1% | $3.45 | 76k | 105 |
| 26 | Kimi K2.7 Code | 30.5% | $2.82 | 59k | 149 |
| 27 | Claude Sonnet 4.6 [high] | 29.9% | $5.52 | 76k | 134 |
| 28 | Gemini 3.1 Pro [high] | 11.7% | $2.14 | 28k | 76 |
“Best per model” shows each model’s highest source-matched Pass@1 in this capture; switch to all effort levels to see every retained configuration. Efforts differ across models, so this view is not a comparison at matched effort. The source was generated September 22 and captured October 1. Small gaps may not be statistically meaningful — confirm on deepswe.datacurve.ai.
When the cheaper one wins
GLM 5.3 is cheaper on output at $4.40 per million tokens against $50 for GPT-6 Astra — about 11×. Use the cheaper tier for classification, extraction, summarisation, and any task where the expensive model’s extra score does not change the accepted output. The expensive one only pays if your hardest task actually fails on the cheap tier. On DeepSWE 1.1, GLM 5.3 is 69% Pass@1 at $3.99/task versus GPT-6 Astra at 74.1% / $4.43/task. This compares standard short-context token rates, not total cost per accepted task. Include input, cache, context tiers, schedules, retries and review in your pilot.
Run the model pickerEvidence confidence: Medium
How strong and complete the evidence behind this comparison is — not a prediction of which model is better.
- Pricing verified on both sides — Input and output rates are verified for both models.
- 4/5 core specs verified on both sides — Not published for at least one side: parameter count.
- No directly aligned benchmark setup — No benchmark has matching cited sources, source dates, units, effort and recorded run details on both models.
- Verified within the last 90 days — Older of the two catalog checks was 25 days ago.
- Both models carry source citations — Each side has at least two catalog sources on record.
Source receipts
Each catalog figure was checked against the provider or an independent second source on the date shown.
- GLM 5.3: Z.ai GLM-5.3 announcement (accessed 2026-10-01)
- GLM 5.3: GLM-5.3 official API limits and account/protocol restrictions (accessed 2026-10-01)
- GLM 5.3: Z.ai exact output ceiling (accessed 2026-10-01)
- GLM 5.3: Z.ai current first-party prices (accessed 2026-10-01)
- GLM 5.3: GLM-5.3 checkpoint and publisher benchmarks (accessed 2026-10-01)
- GLM 5.3: GLM-5.3 exact custom licence (accessed 2026-10-01)
- GPT-6 Astra: OpenAI model card — whole-profile source check October 1 (accessed 2026-10-01)
- GPT-6 Astra: OpenAI current API pricing and processing tiers (accessed 2026-10-01)
- GPT-6 Astra: OpenAI API changelog — September 3 release; checked October 1 (accessed 2026-10-01)
- GPT-6 Astra: OpenAI — Introducing GPT-6 Astra (accessed 2026-09-03)
- GPT-6 Astra: OpenAI API pricing (gpt-6-astra $10/$50) (accessed 2026-09-03)
- GPT-6 Astra: CNBC — OpenAI launches GPT-6 Astra (accessed 2026-09-03)
- GPT-6 Astra: OpenAI model docs — gpt-6-astra (128K max output, effort levels) (accessed 2026-09-05)
- GPT-6 Astra: OpenAI API pricing (gpt-6-astra ≤272K $10/$50, >272K $20/$75) (accessed 2026-09-26)
Related comparisons
- Claude Fable 5.1 vs GLM 5.3
- Claude Fable 5.1 vs GPT-6 Astra
- Claude Fable 5 vs GLM 5.3
- Claude Fable 5 vs GPT-6 Astra
- Claude Mythos 5.1 vs GLM 5.3
- Claude Mythos 5.1 vs GPT-6 Astra
Diving deeper on one model? GLM 5.3 · GPT-6 Astra
Common questions
GLM 5.3 vs GPT-6 Astra
Answered from the verified figures on this page rather than general guidance.
Is GLM 5.3 or GPT-6 Astra cheaper for input?
Is GLM 5.3 or GPT-6 Astra cheaper for output?
Which has the larger context window, GLM 5.3 or GPT-6 Astra?
Do GLM 5.3 and GPT-6 Astra support the same reasoning levels?
Should I use GLM 5.3 or GPT-6 Astra?
Can I self-host GLM 5.3 or GPT-6 Astra?
Next step
Choosing between them
The verified figures that separate this pair, computed from the catalog rather than restated boilerplate.
- Input price: GLM 5.3 costs $1.40 per 1M tokens versus $10 for GPT-6 Astra — a 7.1x difference at the headline tier.
- Context: GPT-6 Astra takes 1.05M against 1M for GLM 5.3 — only decisive if your prompts approach the smaller window.
- Deployment: GLM 5.3 publishes weights you can self-host; the other is API-only.
Choosing a harness rather than a model — Cursor, Copilot, Claude Code, Windsurf, or Lovable? Compare agentic harnesses · Latest releases.