d'AI21 revolutionary hybrid SSM-Transformer combining Mamba state space modèles avec attention layers pour 256K contexte efficacité.
Technical Specifications
- Provider
- AI21
- Category
- flagship
- Release Date
- 2024-08
- Batch API Discount
- 50% off async batch
- Consumer Subscription
- Free / Self-Hosted
- Reasoning Effort Control
- Fixed
- Structured Outputs
- Yes
- Output Modalities
- text
- Official Documentation
- Visit Official Docs ↗
Data Privacy, Licensing & Deployment Governance
Official Provider Release Intelligence, API Quotas & Agentic Specs
- Official API Model ID (SDK String)
- jamba-1-5-large
- Supported SDK & API Protocol
- OpenAI-Compatible REST / vLLM / SGLang / HuggingFace TGI
- Lifecycle & Deprecation Status
- Legacy Stable (Upgrade path available in AI21 lineup)
- Production Rate Limits (RPM & TPM)
- Unlimited (Hardware-bound on self-hosted GPU) / 10,000 RPM Cloud · Unlimited local throughput / 10M+ TPM Managed Cloud
- Long-Context Pricing Tier (>128K / >200K)
- Flat rate ($2/1M in) across full 256K window
- Prompt Cache Write & TTL Policy
- Cached input at $0.5/1M tokens
- Computer Use & Browser GUI Agent
- API / CLI Tool-Use & DOM Agent ready (no raw pixel desktop click loop)
- MCP & Parallel Tool Calling
- Native MCP Client/Server ready · Parallel function calling · 100% Strict JSON Schema
- Web Search Grounding & Code Sandbox
- Supported via external Tavily/Exa Search & E2B/Docker Sandbox tools
- Fine-Tuning & Distillation Support
- Full Weights SFT, LoRA, QLoRA, DPO & GRPO RL (Unsloth / Axolotl / Llama-Factory)
- Multimodal Tokenization Billing
- Text BPE (~4 chars/token)
- Weights Precision & Quantization
- BF16 / FP16, FP8 Dynamic, INT4 AWQ / GPTQ, GGUF (Q4_K_M / Q8_0)
- Minimum GPU VRAM Requirement
- 247 GB VRAM (4-bit Quantized) · 856 GB VRAM (Native BF16)
- Supported Inference Engines
- vLLM, SGLang, Ollama, llama.cpp, TensorRT-LLM, HuggingFace TGI
- System Card & Safety Framework
- Open-Weights System Card + Customizable Guardrail Layer (Llama Guard / ShieldGemma)
- Enterprise Compliance (SOC2 / HIPAA / ZDR)
- 100% Air-Gapped On-Premises / Private VPC · Full HIPAA, GDPR & SOC2 Control
Jamba 1.5 Large Cost, Context, Hardware & API Integration
For single-user chat, a $20/mo subscription caps your bill below this API volume.
Break-even point: ~51 requests/day equals $20/month.
Same Monthly Workload Cost Across Popular Reference Models
Live Prompt Paste Tokenizer & Cost Counter
Paste real system instructions, JSON schemas, or code to measure BPE token counts across tokenizers.
Hybrid Prompt Routing Cascade Simulator
Route routine queries to a fast lightweight model ($0.15/1M) and escalate only complex reasoning turns to Jamba 1.5 Large:
Benchmark Scores & 6-Axis Capability Radar
- SWE-bench Verified
- 41.2%
- GPQA Diamond
- 51%
- MMLU-Pro
- 70.2%
- HumanEval
- 86.5%
- Humanity's Last Exam
- 6.9%
- Terminal-Bench
- 26.5%
- SciCode
- 27.2%
- LiveCodeBench
- 45%
- AIME
- 54%
- Long-Context Reasoning
- 75%
- Chatbot Arena Elo
- 1275pts
Strengths
- Innovative Mamba-Transformer architecture dramatically cuts KV cache mémoire scaling
- True 256K à long contexte ingestion avec supérieur vitesse sur pure attention networks
- à poids ouverts version avec comprehensive d'entreprise outil support
Limitations & Trade-offs
- massif 398B total paramètres require spécialisé multinœud clusters
- moins community écosystème familiarity avec hybrid SSM runtimes
Editorial Assessment
un technological triumph demonstrating le power de Mamba-Transformer architectures à de pointe échelle.
Best for: massif document intelligence, long juridique contract analyse, et hybrid architecture recherche
Compare Jamba 1.5 Large
Frequently Asked Questions about Jamba 1.5 Large
How much does the Jamba 1.5 Large API cost per 1M tokens?+
Jamba 1.5 Large by AI21 is priced at $2 per 1M input tokens and $8 per 1M output tokens. With prompt caching enabled, cached input tokens cost $0.5 per 1M tokens, and asynchronous Batch API requests receive a 50% discount.
Should I use a $20/month subscription or the Jamba 1.5 Large pay-as-you-go API?+
A flat $20/month consumer subscription is predictable for daily interactive chat and manual coding. However, if you are building software, running automated workflows, or sending low-to-moderate volume with prompt caching enabled, the pay-as-you-go API is typically far more cost-effective and offers higher rate limits without manual UI caps.
How many words and pages fit in Jamba 1.5 Large's context window?+
Jamba 1.5 Large features a 256K tokens context window, which equals approximately 192,000 English words or ~512 standard single-spaced pages in a single prompt. Its maximum output generation per turn is 4,096 tokens.
Can I run Jamba 1.5 Large locally on my own GPU (Open Weights)?+
Yes, Jamba 1.5 Large (398B (94B active)) releases open weights that can be self-hosted locally or on private cloud GPUs using vLLM, Ollama, SGLang, or llama.cpp.
Does AI21 train on data sent to the Jamba 1.5 Large API, and is commercial use allowed?+
100% Isolated when self-hosted on private GPUs or VPC. License tier: Open-Weights Research & Commercial License. Full commercial ownership of generated outputs. Supported deployment channels include: Self-Hosted GPU (vLLM / Ollama), Private VPC, or Managed Serverless API.
How does Jamba 1.5 Large perform on coding and reasoning benchmarks?+
Jamba 1.5 Large achieves a composite Intelligence Index of 42/100, scoring 41.2% on SWE-bench Verified, 51% on GPQA Diamond, 70.2% on MMLU-Pro, and 1275 pts on Chatbot Arena Elo.
What is Jamba 1.5 Large best used for?+
Jamba 1.5 Large is best suited for: massif document intelligence, long juridique contract analyse, et hybrid architecture recherche. Assessment: un technological triumph demonstrating le power de Mamba-Transformer architectures à de pointe échelle.
AI21 Family Lineage, Routing Tiers & Upgrade Path
Step down to a lighter sibling for high-volume routing or step up to AI21's flagship reasoning tier:
AI Glossary & Architecture Concepts
View full glossary →Key machine learning concepts and benchmarks relevant to Jamba 1.5 Large:
L'unité textuelle de base traitée par un modèle linguistique, représentant généralement 3 à 4 caractères ou environ 0,75 mot. Par exemple, une phrase courante est découpée en plusieurs jetons.
Une suite de tests standardisés permettant d'évaluer et de comparer objectivement les performances des modèles d'IA sur des tâches ciblées (code avec SWE-bench, mathématiques avec AIME, etc.).
Le nombre maximal de jetons qu'un modèle peut traiter simultanément (entrée et sortie comprises). Une fenêtre de 1 000 000 jetons permet d'analyser environ 750 000 mots en une seule requête.
La capacité d'un LLM à déterminer quand une action externe est requise et à générer des arguments structurés pour invoquer des API, calculatrices ou bases de données avant de poursuivre.
Un standard ouvert créé par Anthropic permettant aux agents d'IA de se connecter de façon uniforme aux systèmes de fichiers locaux, aux environnements de développement et aux API tierces.
Une optimisation tarifaire et technique conservant les états d'attention des préfixes de prompts récurrents, réduisant les coûts d'entrée de 50 à 90 % tout en abaissant la latence.
Les scores de benchmarks, tarifs par token et limites évoluent rapidement. Vérifiez toujours les spécifications actuelles sur la documentation officielle du fournisseur.