Versioned benchmark collection

Space Flight v1

Harness v1 · 24 measured runs across 1 target. Each entry is a single recorded attempt; repeat-run reliability is not established.

Spacecraft approaching a ringed planet in the Space Flight v1 collection
Spacecraft approaching a ringed planet in the Space Flight v1 collection
24Measured runs
1Frozen targets
20Model families
23Configurations
Collection scope

Recorded evidence, without a synthetic score.

Runs share the same frozen target and prompt within each test, with disclosed provider-specific stacks. Costs are recorded estimates as labelled in each ledger, not subscription invoices.

Targets

  • Explorable space-flight game

Model coverage

  • DeepSeek V4 Pro · Max
  • Claude Fable 5 · Max
  • Claude Fable 5 · Medium
  • GLM 5.3 Flash · Max
  • GPT-5.6 Luna · Max
  • GPT-5.6 Luna · xhigh
  • GPT-5.6 Sol · Ultra
  • GPT-5.6 Terra · Ultra
  • GPT-6 Astra · Max
  • Grok 4.7 · xhigh
  • Kimi K3 · Max
  • Claude Opus 5 · Max
  • Qwen3.8 Max Preview · Unreported
  • Qwen3.8 Max · xhigh
  • DeepSeek V4 Flash · Highest
  • DeepSeek V4.1 Flash · Highest
  • Claude Fable 5.1 · Max
  • Gemini 3.8 Flash · Highest
  • GLM 5.3 Flash · Highest
  • GPT-6 Sol · Max
  • MiMo V2.6 Pro · High
  • Claude Opus 5.5 · Unreported
  • Qwen3.8 Flash · Highest

Evidence available

  • 24 measured run ledgers
  • Public browser builds
  • Cost, token, and workflow timing
  • Prompt SHA-256 0cb5a7152dc72f3d423a18887cd82022452072e4f351a9ca36a952f20d06381d
  • Recorded prompt path benchmarks/prompts/space-flight-game-threejs.v1.md

Still missing

  • Browser/hardware environment
  • Local FPS
  • Final capture metadata
  • Minimum blind-vote sample
  • 5 ledgers in this scope do not record a prompt hash. They remain in the historical scope and are not proof of the frozen prompt.
  • 9 related records are listed separately and are not attached to this prompt.
Public receipts

24 measured run ledgers

Open any row for its recorded workflow cost, timing, token usage, stack disclosure, artifact integrity, and explicit evidence gaps.

01DeepSeek V4 ProMax reasoning · Explorable space-flight gamepartial_token_timing_artifact_and_runtime_ledger · FAIL_VISUAL_VALIDITY · Prompt version v1$0.44 API-equivalent usage accounting, not an itemized subscription cash chargerecorded estimate1h 26m 14.4s wall-clockEnd-to-end workflowOpen 02Claude Fable 5Max reasoning · Explorable space-flight gamepartial_token_timing_and_source_ledger · Prompt version v1$151.55 API-equivalent estimate, not a subscription invoicerecorded estimate1:05:19.2 wall-clockEnd-to-end workflowOpen 03Claude Fable 5Medium reasoning · Explorable space-flight gamepartial_token_timing_and_source_ledger · Prompt version v1 · Prompt hash unavailablePrompt hash was not recorded on this historical ledger. It stays in the historical collection scope and is not proof of the frozen prompt.$22.85 Anthropic first-party Claude API, standard global pricingrecorded estimate18:37.7 wall-clockEnd-to-end workflowOpen 04GLM 5.3 FlashMax reasoning · Explorable space-flight gamepartial_multi_turn_token_timing_and_artifact_ledger · Prompt version v1$4.46 UNVERIFIED reference scenario: GLM-5.2/GLM-5.1 official rates applied to GLM-5.3 token counts; not a verified GLM-5.3 charge or subscription marginal-cash invoicerecorded estimate48m 14.1s wall-clockEnd-to-end workflowOpen 05GPT-5.6 LunaMax reasoning · Explorable space-flight gamepartial_token_timing_source_and_build_ledger · Prompt version v1$0.25 API-equivalent estimate, not a subscription invoicerecorded estimate22m 48s wall-clockEnd-to-end workflowOpen 06GPT-5.6 Lunaxhigh reasoning · Explorable space-flight gamepartial_token_timing_and_source_ledger · Prompt version v1 · Prompt hash unavailablePrompt hash was not recorded on this historical ledger. It stays in the historical collection scope and is not proof of the frozen prompt.$0.33 API-equivalent estimate, not a subscription invoicerecorded estimate10:55.9 wall-clockEnd-to-end workflowOpen 07GPT-5.6 SolUltra reasoning · Explorable space-flight gamepartial_token_timing_and_source_ledger · Prompt version v1 · Prompt hash unavailablePrompt hash was not recorded on this historical ledger. It stays in the historical collection scope and is not proof of the frozen prompt.$17.94 user-supplied API-equivalent pre-request snapshotrecorded estimate37:44.2 wall-clockEnd-to-end workflowOpen 08GPT-5.6 TerraUltra reasoning · Explorable space-flight gamepartial_token_timing_and_source_ledger · Prompt version v1 · Prompt hash unavailablePrompt hash was not recorded on this historical ledger. It stays in the historical collection scope and is not proof of the frozen prompt.$1.44 API-equivalent estimate, not a subscription invoicerecorded estimate16:51.4 wall-clockEnd-to-end workflowOpen 09GPT-6 AstraMax reasoning · Explorable space-flight gamepartial_post_task_metrics_snapshot_with_build_and_live_load · Prompt version v1$20.53 Standard API-equivalent estimate from a qualified cumulative root-session snapshot, not a subscription invoicerecorded estimate63m 51.9s (qualified snapshot span) wall-clockEnd-to-end workflowOpen 10Grok 4.7xhigh reasoning · Explorable space-flight gamecomplete_model_internal_judge_passed · PASS — model-internal flight checker and limited independent browser check · Prompt version v1$5.79 Provider-recorded usage estimate, not an itemized subscription cash chargerecorded estimateNot recordedEnd-to-end workflowOpen 11Kimi K3Max reasoning · Explorable space-flight gamepartial_token_timing_and_source_ledger · Prompt version v1 · Prompt hash unavailablePrompt hash was not recorded on this historical ledger. It stays in the historical collection scope and is not proof of the frozen prompt.$3.25 API-equivalent usage accounting, not an itemized subscription cash chargerecorded estimate1:04:11 wall-clockEnd-to-end workflowOpen 12Claude Opus 5Max reasoning · Explorable space-flight gamepartial_token_timing_artifact_build_ledger · Prompt version v1$65.22 First-party Claude API list-price equivalent, not a subscription invoicerecorded estimate57m 55.5s wall-clockEnd-to-end workflowOpen 13Qwen3.8 Max PreviewUnreported reasoning · Explorable space-flight gamepartial_token_timing_runtime_artifact_visual_self_verification_ledger · Prompt version v1$45.89 Qualified third-party NanoGPT API-list-price scenario for the measured token mix; not an Alibaba Token Plan cash chargerecorded estimate2h 4m 17.3s wall-clockEnd-to-end workflowOpen 14Qwen3.8 Max PreviewUnreported reasoning · Explorable space-flight gamepartial_token_timing_runtime_artifact_ledger_orientation_failure · Prompt version v1$6.95 Third-party NanoGPT API-list-price equivalent for the measured token mix; not an Alibaba Token Plan cash chargerecorded estimate38m 43.2s wall-clockEnd-to-end workflowOpen 15Qwen3.8 Maxxhigh reasoning · Explorable space-flight gameartifact_runtime_metrics_ledger_no_blind_evaluation · PASS · Prompt version v1¥49.71 Independently recomputed first-party API-list-price equivalent for the measured token mix; NOT a billed amount. This run used an Alibaba Model Studio token-plan subscription, under which usage is not itemized as cash. (upper bound)recorded upper bound30m 19.1s wall-clockEnd-to-end workflowOpen 16DeepSeek V4 FlashHighest reasoning · Explorable space-flight gamePASS · Prompt version unreported$0.36 Provider-recorded OpenRouter costrecorded estimate1561.4s wall-clockEnd-to-end workflowOpen 17DeepSeek V4.1 FlashHighest reasoning · Explorable space-flight gamecomplete_token_timing_artifact_and_validation_ledger · pass · Prompt version unreported$0.24 Provider-recorded usage estimate, not an itemized subscription cash chargerecorded estimate2744.4s wall-clockEnd-to-end workflowOpen 18Claude Fable 5.1Max reasoning · Explorable space-flight gamepartial_token_timing_artifact_build_and_independent_browser_input_replay_ledger · Prompt version v1$129.75 Official Anthropic API-list-price equivalent, not an itemized Claude Code subscription charge.recorded estimate1h 12m 48.2s wall-clockEnd-to-end workflowOpen 19Gemini 3.8 FlashHighest reasoning · Explorable space-flight gamecomplete_token_timing_artifact_and_validation_ledger · pass_with_visual_caveat · Prompt version unreported$4.30 Provider-recorded usage estimate, not an itemized subscription cash chargerecorded estimate8790.3s wall-clockEnd-to-end workflowOpen 20GLM 5.3 FlashHighest reasoning · Explorable space-flight gamecomplete_token_timing_artifact_and_validation_ledger · pass · Prompt version unreported$1.07 Provider-recorded usage estimate, not an itemized subscription cash chargerecorded estimate11664.2s wall-clockEnd-to-end workflowOpen 21GPT-6 SolMax reasoning · Explorable space-flight gamepartial_asset_quality_gate_unverified_with_playable_stage1 · Prompt version v1$4.59 API-equivalent list price, not a marginal subscription cash chargerecorded estimate2762.1s wall-clockEnd-to-end workflowOpen 22MiMo V2.6 ProHigh reasoning · Explorable space-flight gamecomplete_with_verification_caveats · Prompt version v1$0.29 Provider-recorded usage estimate, not an itemized subscription cash chargerecorded estimate2541.4s wall-clockEnd-to-end workflowOpen 23Claude Opus 5.5Unreported reasoning · Explorable space-flight gamecomplete_with_verification_caveats · Prompt version v1$13.88 API-equivalent list price, not a marginal subscription cash chargerecorded estimateNot recordedEnd-to-end workflowOpen 24Qwen3.8 FlashHighest reasoning · Explorable space-flight gamecomplete_token_timing_artifact_and_validation_ledger · pass · Prompt version unreported$0.87 Provider-recorded usage estimate, not an itemized subscription cash chargerecorded estimate9230.9s wall-clockEnd-to-end workflowOpen
Prompt provenance

Exact prompt receipts

A path is shown only when the published record includes the prompt hash that identifies it. Campaign evidence and projections keep the hash gap visible.

space-flight-game-threejs

SHA-256 0cb5a7152dc72f3d423a18887cd82022452072e4f351a9ca36a952f20d06381d

Path benchmarks/prompts/space-flight-game-threejs.v1.md

Open the released prompt pack
Separate records

Related records with unknown prompt provenance

These published records stay separate from the prompt receipts above. Unknown provenance is why they are listed here.

Limits

Evidence and material gaps

  • Repeat-run reliability evidence is not established.