Quick Take

On June 9, 2026, Cohere and Anthropic released North Mini Code and Claude Fable 5, respectively. These models represent distinct approaches to AI utility: North Mini Code focuses on extreme efficiency and zero-cost accessibility, while Claude Fable 5 positions itself as a high-performance reasoning engine for demanding tasks.

Benchmark Read

Claude Fable 5 consistently outperforms North Mini Code across all tracked metrics. With an Intelligence index of 64.9 compared to 27.6, and a Coding index of 62 versus 33.4, Claude Fable 5 demonstrates significantly higher capability.

Specific benchmark performance highlights the gap:

• GPQA: 0.926 (Fable 5) vs 0.757 (Mini Code)
• HLE: 0.533 (Fable 5) vs 0.099 (Mini Code)
• SciCode: 0.602 (Fable 5) vs 0.382 (Mini Code)
• TAU2: 0.985 (Fable 5) vs 0.374 (Mini Code)
• LCR: 0.7 (Fable 5) vs 0.323 (Mini Code)

Cost and Speed

The models offer vastly different operational profiles. North Mini Code is entirely free, with input and output costs at $0.00/1M tokens. It is optimized for speed, boasting an output rate of 208.137 tok/s and a time to first token of 0.189s.

Conversely, Claude Fable 5 is a premium model. It costs $12.50/1M for input and $50.00/1M for output, resulting in a blended cost of $21.88/1M. Its performance is significantly slower, with an output speed of 63.679 tok/s and a notable time to first token of 54.631s.

Best Fit

North Mini Code is best suited for developers and users who need high-frequency, low-latency coding assistance without incurring costs. Its speed makes it ideal for real-time applications where immediate feedback is required.

Claude Fable 5 is the superior choice for complex, reasoning-heavy workflows. Its high intelligence and coding indices make it appropriate for advanced scientific research, intricate logic puzzles, and tasks where accuracy is more critical than raw speed or cost.