LONG CONTEXT256K CONTEXTTEXTSSM-TRANSFORMERNO CACHE ROW
Jamba Large 1.7 API Pricing
Jamba Large 1.7 is AI21's flagship hybrid SSM-Transformer model, built for long-document reasoning and RAG at a 256K-token context. AI21 Studio lists $2/M input and $8/M output. Pulled from ai21.com.
Input - per 1M tokens
$2.00/M
Source AI21 Studio public list
Output - per 1M tokens
$8.00/M
Context 256K public list
Cache N/A
$2.00/M
Cache no public row not listed
Effective - agentic blend
$2.48/M
92/8 split - no cache discount
§ 01 / TERMINAL
Run the numbers.
Live calculator pre-loaded with Jamba Large 1.7 public token rates. AI21 does not publish a separate prompt-cache token rate for this model.
$ /mo
Workload split
Tokens you can process
—
Words equivalent (English)
—
Effective rate
—
§ 02 / SCENARIOS
Real-world presets.
LONG DOC ANALYSIS
Reading 100-page contracts
$0.424/doc
CODING AGENT
Codebase-scale migration
$0.320/task
RAG - SUPPORT
Ticket triage
$0.024/ticket
ASSISTANT
Research planning turn
$0.052/turn
Estimate · jamba-tokenizer-estimate · ≈3.85 chars/token Auto-counts as you type
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.
Characters 476
Words 77
Tokens (estimated) 124 tokens
Cost as input · uncached $0.00025 USD
Cost as output · uncached $0.00099 USD
| Model | Input /M | Output /M | Effective blended | Context | Best for |
|---|---|---|---|---|---|
| Jamba Large 1.7 Current | $2.00 | $8.00 | $2.48 agentic 92/8 | 256K | Long-document RAG and analysis |
| Jamba2 Mini | $0.20 | $0.40 | $0.216 −11× | 256K | High-volume long-context work |
| Mistral Large 3 | $0.50 cache $0.05 | $1.50 | $0.241 −10× | documented elsewhere — not on pricing page | Open-weight flagship |
| GPT-5.4 | $2.50 cache $0.25 | $15.00 | $1.80 −27% | 1.05M | Frontier reasoning |
| Claude Sonnet 4.6 | $3.00 cache $0.30 | $15.00 | $2.05 −17% | 1M | Frontier coding and agents |
Frequently asked.
Practical Jamba Large 1.7 pricing questions, with AI21's published rates kept separate from workload assumptions.
Q · 01 What is Jamba Large 1.7 priced at? +
On AI21 Studio it lists
$2/M input and $8/M output, with a 256K-token context. Verified on the AI21 pricing page 2026-07-11. It is also available on Amazon Bedrock.Q · 02 Why is Jamba cheaper for long context? +
Jamba uses a hybrid SSM-Transformer (Mamba) architecture whose state-space layers use less memory on long sequences than pure attention, so AI21 can offer a 256K context at a lower price than many Transformer-only long-context models.
Q · 03 Is there a prompt-cache discount? +
No separate cache-read rate is published for Jamba Large, so this calculator keeps cached input equal to base input (
$2/M). The token-only blended cost is about $2.48/M.Q · 04 How does it compare with Jamba Mini? +
AI21's current Mini is Jamba2 Mini, at
$0.20/M input and $0.40/M output - ten times cheaper on input and twenty times on output, with the same 256K context. It replaced Jamba Mini 1.7, retired February 1, 2026, at the same price. Use Mini for high-volume long-context work and Large for the hardest reasoning.Q · 05 Which API model ID should I use? +
AI21's docs list
jamba-large and jamba-large-1.7 as pointing to the dated snapshot jamba-large-1.7-2025-07, and advise pinning dated versions. Re-verify the price on the AI21 pricing page before pinning a production integration.