Models

Language models: price, performance, and what we measured

For each model - official pricing, the benchmarks we ran it through, and separately what vendors report.

Anthropic

Maker of the Claude family and the Claude Code agent. Its models carry most of our day-to-day engineering work.

Claude Fable 5

Anthropic's flagship coding model. In our test it produced the more modular architecture and the better-looking result - but took twice as long.

API pricing: $10 / $50 per million tokens
We tested it on: 2 Runs

View model

OpenAI

Maker of the GPT family and the Codex agent. Publishes its own benchmark tables alongside each launch.

1 model

GPT-5.6 Sol

OpenAI's flagship coding model. In our test it finished in about half the time using ~26% fewer tokens, with a more compact and runtime-efficient implementation.

API pricing: $5 / $30 per million tokens
We tested it on: 1 Run

View model

Stealth (OpenRouter)

An anonymized stealth-preview model served through OpenRouter. The vendor is undisclosed during the preview period.

1 model

An anonymized stealth-preview model on OpenRouter (stealth/ox-alpha). In our first benchmark it built a full app fast and on a fraction of the tokens - but with weaker tenant isolation, no tests, and no automatic UI refresh.

We tested it on: 1 Run

View model

SpaceXAI

Maker of the Grok family. Grok 4.6 was co-developed with Cursor, which makes some launch benchmarks non-independent.

1 model

Grok 4.6

SpaceXAI's low-cost frontier model. Ties GPT-5.6 Sol on the Artificial Analysis intelligence index at a far lower price, but is weaker on long-horizon agent benchmarks. Not yet tested by us.

API pricing: $2 / $6 per million tokens
We tested it on: Not yet tested by us

View model

DevShift benchmarks

Each one is a real use case with a single prompt, live output, and numbers we measured ourselves.

Benchmarks