Metaphor

the most cost efficient coding agent

Next hackathon -

[Metaphor @withmetaphor]

10 days left to register for the Metaphor hackathon.

The rule is simple - team up with your favorite coding agent and build something fun.

AI Credits and snacks are on us.

Register here:
https://t.co/lEkvrnEa1k

[Metaphor @withmetaphor]

@itsaishirl

[Metaphor @withmetaphor]

You and your favorite coding agent. team up.

[Jeet @subhajitsh]

We're hosting the AI-Pilled Builders Hackathon in SF on 21st Nov.

Team up with any coding agents and ship something cool.

$15k in prize pool, winner gets Apple MacMini, and $200+ in credits for everyone.

thanks for sponsoring @smolmachines @freestyle_dev and @concentrateai

Register here

https://t.co/AMIqA2EGDB

[HG @Husain07Husain]

Metaphor now on Windows.

No usage limits for open source models with 60,000 mins of cloud usage

Download here: https://t.co/dxEs8fsk04

@withmetaphor

[Jeet @subhajitsh]

Metaphor is now available on Windows. @withmetaphor

- No usage limits for using Open Models
- 60000 mins of metacloud usage
- work from mobile or desktop app

[Metaphor @withmetaphor]

Metaphor is now available on Windows.

- No usage limits for using Open Models
- 60000 mins of metacloud usage
- work from mobile or desktop app

[Metaphor @withmetaphor]

happy halloween to all devs ♥️

[Jeet @subhajitsh]

this halloween season, we're introducing a spooky theme on @withmetaphor 🎃

Press ⌘ + M to turn it on.

[Jeet @subhajitsh]

The latest Kimi 3.1 Pro and Kimi 3.1 Pro High-Speed models are now available on Metaphor. No usage limits.

[Jeet @subhajitsh]

Today, Metaphor Code crossed 100 million tokens processed.

this is just the start.

[Metaphor @withmetaphor]

@mjdwarner no of course not

[Metaphor @withmetaphor]

Best Open models for coding are available now on Metaphor.
GLM-5.3
DeepSeek V4 Pro
Qwen3 Coder Next
Kimi K3
Kimi K2.5

[Metaphor @withmetaphor]

@mjdwarner big big things.

[Metaphor @withmetaphor]

Introducing Metaphor: The most cost-efficient cloud coding agent.

We outperform Claude Code and Codex on Terminal-Bench 4.0 while achieving a 95% cache token rate.

We use ~25% fewer tokens than Claude Code and Codex.

try out free for a week