Jamba2 Mini API Pricing
Jamba2 Mini is the model AI21 now serves as jamba-mini: a 52B-total, 12B-active hybrid SSM-Transformer released January 8, 2026 with open weights under Apache 2.0. AI21 Studio lists it at $0.20/M input and $0.40/M output with a 256K context - the same price as Jamba Mini 1.7, which AI21 retired on February 1, 2026. Checked by hand against ai21.com and AI21's model docs.
Jamba2 Mini API cost calculator.
Live calculator pre-loaded with Jamba2 Mini's public token rates. AI21 publishes no separate prompt-cache rate, so every input token is billed at the input price.
Jamba2 Mini cost on real workloads.
Reading 100-page contracts
Ticket triage
Invoice field extraction
Enterprise Q&A turn
This is a chars-per-token approximation, not a real tokenizer. Actual tokens vary by language, code density, and tool-call overhead — counts are typically ±10–20% off for English prose, more for code or non-Latin scripts. For exact billing, use the vendor's official tokenizer.
| Model | Input /M | Output /M | Effective blended | Context | Best for |
|---|---|---|---|---|---|
| Jamba2 Mini Current | $0.20 | $0.40 | $0.216 agentic 92/8 | 256K | Cheap 256K-context enterprise Q&A |
| Jamba Large 1.7 | $2.00 | $8.00 | $2.48 +11× | 256K | AI21's larger Jamba for harder reasoning |
| Jamba Mini 1.7 | $0.20 | $0.40 | $0.216 retired | 256K | The Jamba Mini this model replaced |
| Mistral Small 4 | $0.15 cache $0.015 | $0.60 | $0.0842 −2.6× | 256K | Budget model from an independent lab |
| Gemini 3.5 Flash-Lite | $0.30 cache $0.03 | $2.50 | $0.272 +26% | 1M | Budget multimodal alternative |
| Claude Haiku 5.5 | $0.10 cache $0.01 | $0.50 | $0.0682 −3.2× | 1M | Anthropic's fast, high-volume tier |
Jamba2 Mini API pricing FAQ.
Practical Jamba2 Mini pricing questions - the rate, the version behind the alias, and what the page does not include.
Q · 01 What is Jamba2 Mini priced at? +
$0.20/M input and $0.40/M output, with a 256K-token context. The pricing page names the row simply "Jamba Mini"; AI21's model docs identify the version behind it as Jamba2 Mini.Q · 02 Which model ID should I use? +
jamba-mini-2-2026-01 is the dated snapshot; jamba-mini-2 and the unversioned jamba-mini both point to it today, but the unversioned alias moved once already, from Jamba Mini 1.7 in February 2026, and can move again.Q · 03 Did the price change from Jamba Mini 1.7? +
$0.20/$0.40 before AI21 retired it on February 1, 2026, and AI21 has kept its Mini tier at that price since Jamba 1.5. Code that called the jamba-mini alias moved to Jamba2 Mini without a change in per-token cost.Q · 04 Is there a prompt-cache discount? +
$0.20/M. Under our standard 92/8 input/output blend that is $0.216/M.Q · 05 Can I self-host it instead? +
Q · 06 How does it compare with Jamba Large? +
$2.00/M input and $8.00/M output - ten and twenty times this model - with the same 256K context. Large is still the 1.7 generation; AI21 has not published a Jamba2 Large.