IBM · Updated 4 November 2026
Granite 4.2 8B
Granite 4.2 8B from IBM is #50 of 54 models on the Artificial Analysis Intelligence Index, with a score of 5. It costs $0.11 per million tokens ($0.06 in, $0.25 out) and writes 75 tok/s. Its weights are open: anyone can download and run it.
- Intelligence
- 5 · 50th of 54
- Coding
- Not scored yet
- Price per 1M tokens
- $0.11 blended · $0.06 input · $0.25 output
- Output speed
- 75 tok/s
- Time to first token
- 0.4 s
- Context window
- 131K tokens
- Open weights
- Yes
- Released
- 25 August 2026
How it compares
| # | Model | Intelligence | Coding | Price / 1M tokens | Speed | Context |
|---|---|---|---|---|---|---|
| 1 | Granite 4.2 8B IBM · open weights | 5 | — | $0.11 | 75 tok/s | 131K |
| 2 | Gemini 4 Carbon Google | 60 | — | $8.00 | 92 tok/s | — |
| 3 | Gemini 3.1 Flash-Lite Google | 6 | — | $0.56 | — | 1M |
| 4 | Nemotron 3.5 Lightning NVIDIA · open weights | 6 | — | $0.10 | 271 tok/s | 262K |
| 5 | Granite 4.2 3B IBM | 5 | — | — | — | — |
Cheaper alternative within 3 points: Nemotron 3.5 Lightning at $0.10 per million tokens.
Data: Artificial Analysis (intelligence and coding indexes, prices, speed) and OpenRouter (open weights, context windows). Updated 4 November 2026; refreshed every 6 hours.
More rankings
All rankings · Best overall · Best open model · Best value · Cheapest good model · Best for coding · Fastest · Longest context · Price table