Home / Directory / Agencies & professional services / Fine-tuning & custom models / AI21 Labs
AI21 Labs
Foundation-model lab behind the Jamba family, with AI21 Studio APIs, post-training on customer data, and enterprise customization for production language apps. In May 2026 AI21 cut its staff from about 180 to about 70, said it would stop selling standalone models, and refocused on Maestro, its agent-optimization system. The hosted Jamba API, Maestro API and File Library were retired on August 9, 2026, with AI21 Gateway named as the replacement.
Product and platform teams that want Jamba Mini/Large APIs or post-training on their data without adopting a full enterprise agent suite.
Teams that want a turnkey Playbook/agent studio with brand controls as the primary product.
Verdict
AI21 Labs was a good fit when you wanted Jamba models and a clear path to customize on your data through Studio and post-training. Its hosted Jamba API, Maestro API and File Library were retired on August 9, 2026, with AI21 Gateway named as the replacement, so the Pay As You Go rates this review lists (Jamba Mini at $0.20 / $0.40 per 1M tokens and Jamba Large at $2 / $8) no longer apply to that API. Builders liked the long-context efficiency story, and the Jamba models remain available through self-hosting and third-party hosts. It is a model-lab peer to Writer's fuller enterprise agent stack and PremAI's private deployment focus.
Score Breakdown
How AI21 Labs scores in the categories that matter to its buyers.
Pricing
ai21.com/pricing still lists the Pay As You Go rates below, but AI21 retired the hosted Jamba API, Maestro API and File Library on August 9, 2026, so those rates no longer apply to that API. AI21 Gateway, named as the replacement, is free during a beta preview. USD rates as listed in November 2026.
| Plan / SKU | Meter | Price (USD) | What stands out |
|---|---|---|---|
| Free trial | Credits | $10 for 7 days | No credit card required |
| Jamba Mini | Per 1M tokens | $0.20 in / $0.40 out | Efficient lightweight model |
| Jamba Large | Per 1M tokens | $2 in / $8 out | Long-context production model |
| AI21 Gateway | Beta preview | Free | Drop-in endpoint that cuts agent token costs |
| Custom Plan | Volume / private cloud | Sales quote | Discounts, rate limits, dedicated support |
The Field at a Glance
Where AI21 Labs ranks among Fine-tuning & custom models vendors we reviewed, by Overall Score and relative typical engagement cost.
AI21 Labs scores 7.2, behind Writer (8.1) and ahead of PremAI (6.7) among the three Fine-tuning & custom models vendors we reviewed. It plots at the lowest relative typical engagement cost of the three; Writer's Enterprise agent suite lands higher.
Use-case matrix
| Use case | Fit | Notes |
|---|---|---|
| Jamba API inference | Strong | Clear public rates and Studio access. |
| Post-training on proprietary data | Strong | Positioned product path. |
| Long-context document apps | Strong | Jamba efficiency pitch. |
| Full enterprise agent studio | Mixed | Writer is broader on Playbooks/governance. |
| Private GPU fine-tune lab only | Mixed | PremAI may fit private-training-first buyers. |
| No-code brand content desk | Poor | AI21 is developer and API first and does not offer a copy desk. |
Review Excerpts
Excerpts from company materials and public coverage of AI21 Labs, plus representative buyer views drawn from public information. Paid reviews and pay-for-play sites were excluded.
“Jamba Mini made high-volume classification cheap enough that we stopped paying a larger general model for simple routing.”
Representative buyer view
“We prototyped long-context summarization in Studio, then moved steady traffic to pay-as-you-go once the prompts settled.”
Representative buyer view
“AI21 leaves the application layer to you. Evals, guardrails and product UX are still your work.”
Representative buyer view
Where customer reviews couldn't be found, our analysts attempted to give a voice to buyer sentiment, drawing on real information from review sites, forums, and product documentation.
Based on: AI21’s website and Calcalist’s report on the restructuring
Company Facts
- Founded
- 2017 (CB Insights)
- Headquarters
- Tel-Aviv, Israel (CB Insights)
- Total funding raised
- $626.5M (CB Insights)
- Key investors
- Nebius, NVIDIA, Google, Ahren, Intel Capital (CB Insights)
- Website
- ai21.com
Key Leadership
Yoav Shoham
Co-founder and Chairman
Ori Goshen
CEO and Co-Founder
Listed as CEO and co-founder on AI21’s about page. Amnon Shashua left the company in July 2026 (Calcalist).
Methodology
This page is an independent evaluation of AI21 Labs for buyers comparing options in fine-tuning & custom models. AI Industry Reviews accepts no sponsorships, advertising, or pay-for-placement fees. AI21 Labs did not pay for this review.
What we scored
The headline number is an Overall Score on a 0-10 scale. Eight criteria fall under it in two groups.
- Buyer outcomes (for fine-tuning & custom models)
- Foundation model quality (Jamba)
- Customization / post-training path
- API & Studio developer experience
- Enterprise packaging & support
- Company & commercial
- Innovation & product leadership
- Project management & communication
- Pricing
- Contract fairness
Pricing measures whether the price looks fair for the value delivered, including packaging and renewal friction that show up in real buying cycles.
Score Composition
| Input | Weight | What it covers |
|---|---|---|
| Reviews | 40% | A proprietary read of what practitioners say about likes, complaints, and day-to-day use, including public review sites, forums, and private chat rooms. Paid reviews and pay-for-play sites are out of scope. |
| Product | 35% | Hands-on look at screens and workflows. |
| Pricing | 15% | Whether the price looks fair for what you get. |
| Docs & training | 10% | Docs, tutorials, and training material. |
How we balanced the evidence
The Overall Score is the simple average of the eight criteria. Recommendation language follows that score and the use-case matrix above.
Scope
AI21 Labs is graded here as fine-tuning & custom models. Criteria scores can move as more review volume and product checks are added.