AI21 LabsLanguage modelActive
Jamba Large 1.7
Jamba Large 1.7 has no overall index value yet because there are too few independent benchmark results.
Intelligencenot ranked yet
n/aOverall indexhigher is better
Speednot measured
n/ano measurement yet
PriceRank 68 of 95
Input $2 · Output $8
$3.50per 1M tokenslower is better
ContextRank 76 of 102
256Ktokens at once≈ 380 A4 pages
PopularityRank 611 of 875
16out of 100same as last month
Summary
It is fairly expensive: $3.50 per 1M tokens (input and output blended); a typical model costs $1.56.
Jamba Large 1.7 focuses on stronger grounding in supplied context, according to AI21. In the company’s FACTS evaluation, its score rises from 0.758 in version 1.6 to 0.832. The hybrid SSM-Transformer architecture handles 256K context tokens.
AI21 lists 398 billion total parameters and 94 billion active parameters. The weights use the Jamba Open Model License, with Hugging Face access requiring acceptance of the terms and sharing contact information.
All benchmark results
Every published result with its source, test setup and comparison models.
Show all 13 published benchmark results
Page 1 of 2
Measurements without matching peer values
These measurements have no matching peer values under the same test conditions. Their original values and sources remain available here.
CritPt v2: Overall (Show measurement, test conditions, and source)
- Source value
- 0
- Score
- 0
- Metric
- pass@1 accuracy
- Unit
- %
- Benchmark version
- v2
- Category
- reasoning
- Task
- overall
- Task label
- Overall
- Direction
- Higher is better
- Source type
- independent-evaluator
- Evaluator
- Artificial Analysis
- Status
- disputed
- Retrieved at
- 2026-11-02
- Methodology
- Exact AA configuration: Jamba 1.7 Large (release jamba-1-7-large, row jamba-1-7-large, UUID 78452c64-7303-4192-bca5-2d9ec5c623d4, reasoning=false, effort=not specified, suffix=not specified). AA marks this model variant as non-reasoning and reports 0 reasoning tokens. The full CritPt v2 run covers 70 questions with five repeats. The official grading server reports pass@1.. Source model ID: artificial-analysis/78452c64-7303-4192-bca5-2d9ec5c623d4/jamba-1-7-large/jamba-1-7-large. Source model name: Jamba Large 1.7. Provider reported by source: AI21 Labs. Model snapshot: Jamba 1.7 Large. Evaluator: Artificial Analysis. Dataset type: artificial-analysis-evaluation. Tool policy: Python functions, symbolic expressions, and numerical answers; official grading server. Attempts: 70 questions, five repeats per question, pass@1. Reasoning: non-reasoning. Reasoning effort: none. Score display: 0.0%. Source ID: artificial-analysis. Evidence capture ID: aa-model-page-2026-11-02-67d46be20a5b. Source payload hash: 67d46be20a5b412db6cfb60ad190441fcaa5513500b8fd47d033168d0e97b6b2. Parser version: artificial-analysis-rsc-models-v3. Imported at: 2026-11-02. Source cohort: artificial-analysis-critpt-v2:v2:standalone:non-reasoning:effort=unspecified:suffix=unspecified
- Comparison methodology
- Exact AA configuration: Jamba 1.7 Large (release jamba-1-7-large, row jamba-1-7-large, UUID 78452c64-7303-4192-bca5-2d9ec5c623d4, reasoning=false, effort=not specified, suffix=not specified). AA marks this model variant as non-reasoning and reports 0 reasoning tokens. The full CritPt v2 run covers 70 questions with five repeats. The official grading server reports pass@1.. Tool policy: Python functions, symbolic expressions, and numerical answers; official grading server. Attempts: 70 questions, five repeats per question, pass@1. Reasoning: non-reasoning. Reasoning effort: none
- Context
- This AA source field contains the original CritPt score. The earlier CritPt v2 attribution was incorrect; the value and former label remain as disputed historical evidence.. The source dataset marks this result as disputed. It remains visible as documented evidence, but is not reliable comparative proof.
IFBench: Overall (Show measurement, test conditions, and source)
- Source value
- 35.1700680272109
- Score
- 35.1700680272109
- Metric
- IFBench prompt-level loose accuracy (pass@1)
- Unit
- %
- Benchmark version
- AA test set: 294 questions
- Category
- general
- Task
- overall
- Task label
- Overall
- Direction
- Higher is better
- Source type
- independent-evaluator
- Evaluator
- Artificial Analysis
- Status
- superseded
- Retrieved at
- 2026-11-02
- Methodology
- Exact AA configuration: Jamba 1.7 Large (release jamba-1-7-large, row jamba-1-7-large, UUID 78452c64-7303-4192-bca5-2d9ec5c623d4, reasoning=false, effort=not specified, suffix=not specified). AA marks this model variant as non-reasoning and reports 0 reasoning tokens. Artificial Analysis runs 294 questions from the IFBench test set with five repeats per question. The official evaluator scores prompt-level accuracy in loose mode as pass@1.. Source model ID: artificial-analysis/78452c64-7303-4192-bca5-2d9ec5c623d4/jamba-1-7-large/jamba-1-7-large. Source model name: Jamba Large 1.7. Provider reported by source: AI21 Labs. Model snapshot: Jamba 1.7 Large. Evaluator: Artificial Analysis. Dataset type: artificial-analysis-evaluation. Tool policy: Single-turn answer evaluated with the official IFBench source code. Attempts: pass@1, 5 repeats per question. Reasoning: non-reasoning. Reasoning effort: none. Score display: 35.2%. Source ID: artificial-analysis. Evidence capture ID: aa-model-page-2026-11-02-67d46be20a5b. Source payload hash: 67d46be20a5b412db6cfb60ad190441fcaa5513500b8fd47d033168d0e97b6b2. Parser version: artificial-analysis-rsc-models-v3. Imported at: 2026-11-02. Source cohort: artificial-analysis-ifbench:AA test set: 294 questions:standalone:non-reasoning:effort=unspecified:suffix=unspecified
- Comparison methodology
- Exact AA configuration: Jamba 1.7 Large (release jamba-1-7-large, row jamba-1-7-large, UUID 78452c64-7303-4192-bca5-2d9ec5c623d4, reasoning=false, effort=not specified, suffix=not specified). AA marks this model variant as non-reasoning and reports 0 reasoning tokens. Artificial Analysis runs 294 questions from the IFBench test set with five repeats per question. The official evaluator scores prompt-level accuracy in loose mode as pass@1.. Tool policy: Single-turn answer evaluated with the official IFBench source code. Attempts: pass@1, 5 repeats per question. Reasoning: non-reasoning. Reasoning effort: none
- Context
- The often cited 58 refers to verifiable out-of-domain constraint types, not the number of questions in AA's test run.. This result is retained as historical evidence and is not interchangeable with current results.
IFBench: Overall (Show measurement, test conditions, and source)
- Source value
- 35.1700680272109
- Score
- 35.1700680272109
- Metric
- IFBench prompt-level loose accuracy (pass@1)
- Unit
- %
- Benchmark version
- AA test set: 294 questions
- Category
- general
- Task
- overall
- Task label
- Overall
- Direction
- Higher is better
- Source type
- independent-evaluator
- Evaluator
- Artificial Analysis
- Status
- active
- Retrieved at
- 2026-11-06
- Methodology
- Exact AA configuration: Jamba 1.7 Large (release jamba-1-7-large, row jamba-1-7-large, UUID 78452c64-7303-4192-bca5-2d9ec5c623d4, reasoning=false, effort=not specified, suffix=not specified). This source capture does not provide an average reasoning-token count. Artificial Analysis runs 294 questions from the IFBench test set with five repeats per question. The official evaluator scores prompt-level accuracy in loose mode as pass@1.. Source model ID: artificial-analysis/78452c64-7303-4192-bca5-2d9ec5c623d4/jamba-1-7-large/jamba-1-7-large. Source model name: Jamba Large 1.7. Provider reported by source: AI21 Labs. Model snapshot: Jamba 1.7 Large. Evaluator: Artificial Analysis. Dataset type: artificial-analysis-evaluation. Tool policy: Single-turn answer evaluated with the official IFBench source code. Attempts: pass@1, 5 repeats per question. Reasoning: non-reasoning. Reasoning effort: none. Score display: 35.2%. Source ID: artificial-analysis. Evidence capture ID: aa-model-page-2026-11-06-ec1194e008b3. Source payload hash: ec1194e008b3638ef60dbf5249daf9e6047364e22905e745a4af64c5f3f080fe. Parser version: artificial-analysis-rsc-models-v4. Imported at: 2026-11-06. Source cohort: artificial-analysis-ifbench:AA test set: 294 questions:standalone:non-reasoning:effort=unspecified:suffix=unspecified
- Comparison methodology
- Exact AA configuration: Jamba 1.7 Large (release jamba-1-7-large, row jamba-1-7-large, UUID 78452c64-7303-4192-bca5-2d9ec5c623d4, reasoning=false, effort=not specified, suffix=not specified). This source capture does not provide an average reasoning-token count. Artificial Analysis runs 294 questions from the IFBench test set with five repeats per question. The official evaluator scores prompt-level accuracy in loose mode as pass@1.. Tool policy: Single-turn answer evaluated with the official IFBench source code. Attempts: pass@1, 5 repeats per question. Reasoning: non-reasoning. Reasoning effort: none
- Context
- The often cited 58 refers to verifiable out-of-domain constraint types, not the number of questions in AA's test run.
Vectara hallucination rate (Show measurement, test conditions, and source)
- Source value
- 9.7
- Score
- 9.7
- Metric
- Vectara hallucination rate
- Unit
- %
- Category
- hallucination
- Direction
- Lower is better
- Source type
- independent-evaluator
- Evaluator
- Vectara Hallucination Leaderboard
- Status
- active
- Retrieved at
- 2026-10-04
- Methodology
- Independent evaluation. ai21labs/jamba-large-1.7-2025-07; fixed article-summary prompt, commercial HHEM threshold <0.5; answered summaries only; answer rate 98.9%.
- Context
- Automated summary evaluation, not a general factual-accuracy measure. Exact row sample count, run date, and reasoning setting are unspecified. The complementary consistency rate describes the same evaluation, not a second test.
Vectara factual consistency rate (Show measurement, test conditions, and source)
- Source value
- 90.3
- Score
- 90.3
- Metric
- Vectara factual consistency rate
- Unit
- %
- Category
- hallucination
- Direction
- Higher is better
- Source type
- independent-evaluator
- Evaluator
- Vectara Hallucination Leaderboard
- Status
- active
- Retrieved at
- 2026-10-04
- Methodology
- Independent evaluation. ai21labs/jamba-large-1.7-2025-07; same HHEM article-summary evaluation as the hallucination rate; answer rate 98.9%.
- Context
- Complementary result from the same summaries, not independent corroboration or a general knowledge score. Exact run date, sample count, and reasoning setting are unspecified.
Kagi Offline Benchmark (Show measurement, test conditions, and source)
- Source value
- 26.1
- Score
- 26.1
- Metric
- Accuracy (%)
- Unit
- %
- Benchmark version
- Kagi task-list revision 2026-02-13; current page observed 2026-10-17
- Category
- general-reasoning
- Direction
- Higher is better
- Source type
- independent-evaluator
- Evaluator
- Kagi LLM Benchmarking Project
- Status
- active
- Retrieved at
- 2026-10-17
- Methodology
- Independent evaluation. Kagi Offline Benchmark without tools. Exact listed route: ai21/jamba-large-1.7, provider OpenRouter. Kagi says its task list was last revised 2026-02-13 and the test favors models that use chain of thought heavily. The source does not state this row's immutable served checkpoint, reasoning setting, per-model sample count, confidence interval, or execution date.
- Context
- This is a route-level result, not a claim about a pinned Jamba Large 1.7 weight or snapshot. Unknown configuration and sample details remain unpublished.
Kobayashi v0.3.0 core English: autonomous lethal action rate (Show measurement, test conditions, and source)
- Source value
- 80
- Score
- 80
- Metric
- Autonomous lethal action rate
- Unit
- %
- Benchmark version
- Kobayashi v0.3.0, core protocol, English track
- Category
- safety-behavior
- Direction
- Lower is better
- Sample
- 20
- Source type
- independent-evaluator
- Evaluator
- Kobayashi Benchmark run audit
- Status
- active
- Retrieved at
- 2026-10-17
- Methodology
- Independent evaluation. Exact source model ID ai21/jamba-large-1.7; run 20260715T204250Z-ai21-jamba-large-1.7-5acb459a (2026-07-15 20:42:50 UTC); v0.3.0 core protocol, English track; 20 prompts, 20/20 scored; three independent Ollama Cloud judges. The page does not identify the model inference route or an immutable serving revision.
- Context
- The publisher labels this a provisional run; the scorer has not been calibrated against human annotation. Text responses do not establish real-world behavior. The benchmark authors' economic or institutional relationship to AI21 is not stated by the source.
FACTS (Show measurement, test conditions, and source)
- Source value
- 0.832
- Score
- 0.832
- Metric
- FACTS
- Unit
- No unit provided
- Category
- hallucination
- Direction
- Higher is better
- Source type
- vendor-reported
- Evaluator
- AI21 Jamba Large 1.7 model card
- Status
- active
- Retrieved at
- 2026-10-04
- Methodology
- Vendor-reported result. Exact Jamba Large 1.7 column in AI21's Grounding and instruction-following improvements table; FACTS grounding result. Benchmark-specific serving and evaluation settings are not documented.
- Context
- Vendor-reported raw score with an unspecified scale, not normalized to a percentage. Exact benchmark version, harness, sample count, sampling, reasoning setting, and run date are not supplied. Inference examples are not the benchmark protocol.
IFEcal (Show measurement, test conditions, and source)
- Source value
- 0.84
- Score
- 0.84
- Metric
- IFEcal
- Unit
- No unit provided
- Category
- instruction-following
- Direction
- Higher is better
- Source type
- vendor-reported
- Evaluator
- AI21 Jamba Large 1.7 model card
- Status
- active
- Retrieved at
- 2026-10-04
- Methodology
- Vendor-reported result. Exact Jamba Large 1.7 column in AI21's Grounding and instruction-following improvements table; steerability result labelled IFEcal in the source. Benchmark-specific serving and evaluation settings are not documented.
- Context
- Vendor-reported raw score with an unspecified scale. The source spells the benchmark IFEcal; no equivalence to IFEval is established. Exact version, harness, sample count, sampling, reasoning setting, and run date are not supplied. Inference examples are not the benchmark protocol.
VikiKaynak DSSI pipeline style BLEU (Show measurement, test conditions, and source)
- Source value
- 80.5
- Score
- 80.5
- Metric
- BLEU style score, reported out of 100
- Unit
- points
- Benchmark version
- VikiKaynak DSSI Table 7, SIGTURK 2026
- Category
- text generation
- Direction
- Higher is better
- Source type
- independent-evaluator
- Evaluator
- VikiKaynak DSSI evaluation paper, SIGTURK 2026
- Status
- active
- Retrieved at
- 2026-10-17
- Methodology
- Independent evaluation. SIGTURK 2026 Table 7 evaluates Turkish style transfer on Vikikaynak. This is the DSSI pipeline configuration; BLEU measures stylistic similarity. The paper reports BLEU and BERT out of 100. The model-specific scored denominator and full generation settings are not stated.
VikiKaynak zero-shot baseline style BLEU (Show measurement, test conditions, and source)
- Source value
- 75.5
- Score
- 75.5
- Metric
- BLEU style score, reported out of 100
- Unit
- points
- Benchmark version
- VikiKaynak DSSI Table 7, SIGTURK 2026
- Category
- text generation
- Direction
- Higher is better
- Source type
- independent-evaluator
- Evaluator
- VikiKaynak DSSI evaluation paper, SIGTURK 2026
- Status
- active
- Retrieved at
- 2026-10-17
- Methodology
- Independent evaluation. SIGTURK 2026 Table 7 evaluates Turkish style transfer on Vikikaynak. This is the zero-shot baseline configuration using standard prompting; BLEU measures stylistic similarity. The paper reports BLEU and BERT out of 100. The model-specific scored denominator and full generation settings are not stated.
VikiKaynak DSSI pipeline content BERT (Show measurement, test conditions, and source)
- Source value
- 97.6
- Score
- 97.6
- Metric
- BERT content preservation score, reported out of 100
- Unit
- points
- Benchmark version
- VikiKaynak DSSI Table 7, SIGTURK 2026
- Category
- text generation
- Direction
- Higher is better
- Source type
- independent-evaluator
- Evaluator
- VikiKaynak DSSI evaluation paper, SIGTURK 2026
- Status
- active
- Retrieved at
- 2026-10-17
- Methodology
- Independent evaluation. SIGTURK 2026 Table 7 evaluates Turkish style transfer on Vikikaynak. This is the DSSI pipeline configuration; BERT measures content preservation. The paper reports BLEU and BERT out of 100. The model-specific scored denominator and full generation settings are not stated.
Page 1 of 2
Popularity compared
How often the model is searched for, used and mentioned, compared with every model of its category.
Popularity index
The 100 most popular models of this category, 0 to 100, higher is more popular
20 of 101 models
Gemini 4 Argon, Google DeepMind: 92
GPT-6 Astra, OpenAI: 78
Claude Haiku 5.5, Anthropic: 77
Qwen3.5 4B, Alibaba: 76
GPT-5, OpenAI: 73
GPT-6 Sol, OpenAI: 72
Qwen3.5 0.8B, Alibaba: 72
Qwen3.5 2B, Alibaba: 71
Qwen 3.8 27B, Alibaba: 70
Claude Opus 5.5, Anthropic: 70
DeepSeek V4.1 Flash 1017, DeepSeek: 68
Qwen 3.8 Max 0902, Alibaba: 68
Kimi K3, Moonshot AI: 68
DeepSeek-V4-Flash, DeepSeek: 65
GPT-5.6 Terra, OpenAI: 64
Claude Sonnet 5, Anthropic: 63
DiffusionGemma 26B A4B IT, Google: 63
Claude Opus 5, Anthropic: 62
Qwen 3.8 Flash, Alibaba: 62
Jamba Large 1.7, AI21 Labs: 16
As of November 2026 · Source: gradually.ai popularity index
Over time
Weekly values from October 18, 2026 to November 5, 2026
Jamba Large 1.7Sonar ProJamba LargeMistral Small 3.1 24BMinistral 8B
All specifications
Architecture
Input and output
Watermarking and provenance
AI watermarking refers to invisible signals embedded in the output. Visible marks and provenance metadata are listed separately. Unknown means there is insufficient evidence for that particular field.
- AI watermarking
- Unknown
- no source
- Watermarking technique
- Not known
- no source
Licence and access
API
Further details
- Notes
- Distinct 1.7 checkpoint. The undated jamba-large API endpoint points to jamba-large-1.7-2025-07; the older 1.5 and 1.6 versions remain separate.
- Source: Notes
Prices
1 offer, prices in US dollars per 1M tokens, blended from 3 parts input and 1 part output.
| Provider | Input | Output | Blended |
|---|---|---|---|
| AI21 Labsdirect | $2 | $8 | $3.50 |
Direct price according to the developer, checked Oct 29, 2026. Open the LLM price index
More price tiers and rules
- Source pricing conditions
- Input $2, output $8 per 1M tokens (retrieved 2026-10-29)Source
Unchanged since tracking began on Nov 1, 2026.
InputOutput
Frequently asked questions about Jamba Large 1.7
How much does Jamba Large 1.7 cost?
Through the API $2 per 1M input tokens and $8 output. That is fairly expensive.
Prices from Nov 6, 2026
Nov 1, 2026Nov 6, 2026
Unchanged since tracking began on Nov 1, 2026.
How large is the context window of Jamba Large 1.7?
256,000 tokens. That is how much text Jamba Large 1.7 reads at once, about 380 A4 pages.
Page count as a rule of thumb from Artificial Analysis
Kimi K2.6262K
Qwen 3.8 27B262K
Kimi K2.7 Code262K
Kimi K2.5262K
Kimi K2 Thinking262K
Mistral Medium 3.5262K
Jamba Large 1.7Rank 76 of 102256K
Can I run Jamba Large 1.7 locally?
Yes, the weights are open. With 398B parameters it needs a lot of memory, more a server than a laptop.
Stated by AI21 Labs
- Open weights
- yes
- License
- Jamba Open Model License
- Parameters
- 398B
- active per token
- 94B
Where can I use Jamba Large 1.7?
Through the API. Subscription access is not documented in the available plan data.
Through the API
$2 input, $8 output per 1M tokens, unchanged since Nov 1, 2026.
Sources and data date
Show all 22 sources
Benchmark sources at a glance
Which independent evaluators tested this model. This is coverage, not a quality ranking.
| Evaluator | Test series | Composite indices | Recorded active measurements |
|---|---|---|---|
| Artificial Analysis | 1 | 0 | 1 |
Without vendor claims and historical or disputed results; those stay labelled in the single values.
| Type | Evidence and data date |
|---|---|
| Price | Base API prices without caching or batch discounts. Higher context tiers and other rates are listed under Costs. |
| Input | Documented modalities. Retrieved October 29, 2026. |
| Output | Documented modalities. Retrieved October 29, 2026. |
| Research status | Research date October 4, 2026. 15 source URLs checked. This documents the inspected sources, not an exhaustive inventory of every publication. |
| Not published | evaluation-protocol Note The paper does not publish the model-specific scored denominator or full generation settings. |
| Not published | exact-served-checkpoint-and-evaluation-configuration Note Kagi publishes the OpenRouter route name and aggregate accuracy, but not the exact served checkpoint, per-model sample count, confidence interval, reasoning setting, or result date. |
| Additional source | AI21 Jamba Large 1.7 model card (retrieved October 25, 2026) · AI21 Jamba Large 1.7 model card · Editorial description reviewed October 4, 2026 |
| Additional source | Artificial Analysis (retrieved November 2, 2026) |
| Additional source | Artificial Analysis (retrieved November 2, 2026; November 6, 2026) |
| Additional source | Vectara Hallucination Leaderboard (retrieved October 4, 2026; October 18, 2026) |
| Additional source | AI21 Jamba Large 1.7 model card (retrieved October 4, 2026) |
| Additional source | VikiKaynak DSSI evaluation paper, SIGTURK 2026 (retrieved October 17, 2026) |
| Additional source | Kagi LLM Benchmarking Project (retrieved October 17, 2026) |
| Additional source | Kobayashi Benchmark run audit (retrieved October 17, 2026) |
| Additional source | Model metadata (retrieved October 4, 2026)12 |
| Additional source | Model metadata (retrieved October 17, 2026)123 |
| Additional source | Model metadata (retrieved October 18, 2026) |
| Additional source | OpenRouter: Popularity index (derived), CC BY 4.0 |
| Additional source | Vercel AI Gateway: Popularity index (derived), CC BY 4.0 |
| Additional source | Ollama: cumulative downloads for the popularity index |
| Additional source | Replicate: cumulative runs for the popularity index |
| Additional source | Google Trending Now: limited feed selection for Germany and the US |
| Additional source | Creative Commons Attribution 4.0 |