Round 1 · Thu, Oct 8, 2026 · Animation
A bird that can fly
Claude Code and Codex CLI got the same strict prompt: one 2D canvas, one side-view bird, sky only, left to right in about 6 seconds. Both builds passed the automated rules check.
Result: No winner
- Claude Opus 5.50 votes
- GPT-6 Astra0 votes
Voting has closed: nobody voted, so there's no winner.
Which result do you think is best?
The community vote decides this battle's result. Voting has closed: nobody voted, so there's no winner.
No votes yet.
- Claude Opus 5.50% · 0
- GPT-6 Astra0% · 0
Play it yourself
What each of the 2 models built, live in your browser
- Time
- 1m 26s
- Cost
- $0.25est.
- Tests
- Not available
- Tokens
- 57,446
Rules check: passed. One file; 2D canvas only (no WebGL, SVG, CSS animation, libraries or network requests); one canvas that fills and resizes with the window; no text; no errors.
- A small side-view bird with flapping wings flies left to right with a gentle bob and re-enters from the left.
- Deep-blue gradient sky with a glowing sun and drifting clouds.
- Finished in 1 m 26 s.
Run record
- Requested model
- claude-opus-5-5
- Resolved version
- claude-opus-5-5
- Tool / harness
- Claude Code 2.1.294
- Settings
- effort: high
- Attempts
- 1
- Human help
- None recorded
- Tool access
- Read, Write, Edit
- Environment
- Linux server, empty working folder
- Limits
- 15-minute timeout, one attempt
- Input / output tokens
- 49,140 / 8,306
- Token details
- uncached_input: 6 · thinking_output: 3497 · cache_read_input: 39791 · cache_creation_input: 9343 · cache_creation_1h_input: 9343 · cache_creation_5m_input: 0
- Cost basis
- Estimated
- Measurement notes
- Wall time measured by the battle harness (same method for both). Tokens from Claude Code's JSON result. Subscription run. Claude Code also reported a list-price estimate ($0.25); it is not shown because Codex reports no cost, so cost is left out for both.
- Executed
- Oct 8, 2026, 8:06 PM GMT+9
- Time
- 2m 13s
- Cost
- $0.32est.
- Tests
- Not available
- Tokens
- 70,869
Rules check: passed. One file; 2D canvas only (no WebGL, SVG, CSS animation, libraries or network requests); one canvas that fills and resizes with the window; no text; no errors.
- A large, detailed side-view bird (layered wing feathers, pale belly) flaps across from left to right and re-enters from the left.
- Soft pastel sky with a glowing sun and pale clouds.
- Finished in 2 m 13 s.
Run record
- Requested model
- gpt-6-astra
- Resolved version
- Not available
- Tool / harness
- Codex CLI 0.161.0
- Settings
- reasoning_effort: high
- Attempts
- 1
- Human help
- None recorded
- Tool access
- Shell commands in a workspace-write sandbox (network off)
- Environment
- Linux server, empty working folder
- Limits
- 15-minute timeout, one attempt
- Input / output tokens
- 66,986 / 3,883
- Token details
- cached_input: 59904 · reasoning_output: 299
- Cost basis
- Estimated
- Measurement notes
- Wall time measured by the battle harness (same method for both). Tokens from codex exec --json (turn.completed usage). Subscription run via ChatGPT sign-in; Codex reports no per-run cost.
- Executed
- Oct 8, 2026, 8:06 PM GMT+9
How the result is decided
No votes were cast
Both builds followed every rule, so the comparison comes down to craft. The clearest visible differences are the bird's size and detail (Codex's bird is much larger, with layered feathers; Claude's is smaller and simpler) and the sky (deep blue with bright clouds versus a soft pastel palette).
The checks, time and cost are facts about each build, not a score. One observed run under stated conditions — not a universal ranking. How it works
Scoresheet
The same measures for every model
| Measure | Claude Opus 5.5 | GPT-6 Astra |
|---|---|---|
| Result | Unscored | Unscored |
| Run | Completed | Completed |
| Task completion | Complete | Complete |
| Works? | Works | Works |
| Tests passed | Not available | Not available |
| Time | 1m 26s | 2m 13s |
| Cost | $0.25at list price | $0.32at list price |
| Tokens | 57,446 | 70,869 |
| Tool / harness | Claude Code | Codex CLI |
| Attempts | 1 | 1 |
How the cost is calculated
Estimated API cost: the tokens each model used × its provider's official list price per million tokens. Reasoning tokens are billed as output. Discounts and subscriptions are not included (these runs were paid for through subscriptions).
| Fresh input | 6 | × $4/M | $0.0000 |
|---|---|---|---|
| Cache reads | 39,791 | × $0.2/M | $0.0080 |
| Cache writes (1 hour) | 9,343 | × $8/M | $0.07 |
| Output (incl. reasoning) | 8,306 | × $20/M | $0.17 |
| Fresh input | 7,082 | × $10/M | $0.07 |
|---|---|---|---|
| Cache reads | 59,904 | × $1/M | $0.06 |
| Output (incl. reasoning) | 3,883 | × $50/M | $0.19 |
Prices: platform.claude.com (checked 2026-10-10) · developers.openai.com (checked 2026-10-10)
The prompt
Every model received exactly this
Create a bird that can fly.
Follow every rule exactly:
1. Output: exactly one file named index.html in the current directory. Create no other files.
2. Technology: plain HTML and JavaScript drawing on a single 2D canvas (getContext("2d")). No WebGL, no SVG, no CSS animations, no libraries, and no external resources of any kind (no fonts, images or network requests).
3. Canvas: fills the whole browser window and resizes with it. No margins, no scrollbars.
4. Scene: a sky background (a gradient is fine) with at most a sun and some clouds. No ground, no text, no titles, no buttons, no UI of any kind.
5. Bird: exactly one bird, seen from the side, drawn with canvas paths (no emoji or text characters). Its wings must visibly flap.
6. Motion: the bird flies continuously from left to right with a gentle up-and-down motion. One crossing of the screen takes about 6 seconds; after leaving on the right it re-enters from the left.
7. Timing: animate with requestAnimationFrame and use its timestamp so the speed does not depend on the frame rate. Start immediately on load, never require any input, and keep running for as long as the page is open.
8. Do not ask questions; make reasonable choices within these rules.Run details: success criteria, conditions and notes
What counts as success
- Follows every rule in the prompt
- A recognisable bird seen from the side
- Wings visibly flap
- Smooth left-to-right flight with a gentle up-and-down motion
Conditions
- Time limit
- 15 minutes
- Starting materials
- An empty folder
- Network access
- Blocked
- Tools
- Claude Code: Read, Write and Edit tools only; no plugins, MCP servers or skills. Codex CLI: codex exec in a workspace-write sandbox. Both used their own default system prompts.
- Human assistance
- None. No edits after the prompt.
- Attempts
- One attempt per model
- Other rules
- Automated rules check after the run: one file only; 2D canvas only (no WebGL, SVG, CSS animation, libraries or network requests); one canvas that fills and resizes with the window; no text or scrollbars; no errors. Both passed.
Prompt fingerprint (SHA-256): 83c3fd88435d475820f69ed13ff38582700ba6ca1de475006128e272ff08f8ea
More rounds
#8
A tree growing through the four seasons
· 3D scene
- Claude Opus 5.5Unscored
- GPT-6 AstraUnscored
Result: Result pending
#9
A jellyfish drifting through a dark ocean
· 3D scene
- Claude Opus 5.5Unscored
- GPT-6 AstraUnscored
Result: Result pending
#7
A firefly-lit forest at dusk
· 3D scene
- Claude Opus 5.5Unscored
- GPT-6 AstraUnscored
Result: Result pending