J2 Ultra
j2-ultraPricing
List prices at AI21 Studio, in US dollars per million tokens.
| Tier | Input / 1M | Output / 1M | |
|---|---|---|---|
| Standard | $15.00 | $15.00 |
What it costs a month
At AI21 Studio list prices, 30 days.
- Chat assistant5,000 requests a day, 2K in / 0.5K out$5,625
- RAG over documents2,000 requests a day, 8K in / 0.6K out$7,740
- Coding agent1,000 requests a day, 40K in / 2K out$18,900
- Overnight batch20,000 requests a day, 4K in / 0.4K out$39,600
Where to run it
2 hosts serve J2 Ultra. The comparison is on blended price, 3 parts input to 1 part output.
| Host | In / out per 1M | Model id | |
|---|---|---|---|
AI21 StudioLab API j2-ultra | $15.00 / $15.00 | j2-ultra | |
Amazon BedrockCloud ai21.j2-ultra-v1 | $18.80 / $18.80 +25% vs headline | ai21.j2-ultra-v1 |
Capabilities
Core
- Streaming
- Function calling
- JSON mode
- Structured outputs
Multimodal
- Vision
- Image input
- Audio
- Image generation
- Video
Advanced
- Reasoning
- Tool calling
- MCP
- Prompt caching
- Batch API
- Embeddings
- Takes in
- Text
- Produces
- Text
Cheaper alternatives
Current models that cost less and keep most of what this one can do: tools, vision, reasoning, structured output, caching and enough context.
Same model, cheaper host
No host sells it for less than AI21 Studio.
Smaller AI21 Labs models
- 33% less
- 80% less
Other frontier labs
- 47% less
- 68% less
- 73% less
- 80% less
Open-weight picks
- 71% less
- 73% less
- 80% less
- 87% less
Call it
Using AI21 Studio's API.
from litellm import completion
response = completion(
model="ai21/j2-ultra",
messages=[{"role": "user", "content": "Hello"}],
)
print(response.choices[0].message.content)Quick answers
How much does J2 Ultra cost?
On AI21 Studio, J2 Ultra costs $15.00 per million input tokens and $15.00 per million output tokens.
Which APIs offer J2 Ultra?
J2 Ultra runs on 2 APIs: AI21 Studio and Amazon Bedrock.
What is J2 Ultra's context window?
8K tokens, with up to 8K tokens in one reply.
Is J2 Ultra being retired?
No retirement date is scheduled for it on any API we track.
What are cheaper alternatives to J2 Ultra?
Models that cost less and keep most of what it can do: J2 Mid on AI21 Studio (33% less), Claude Opus 5.5 on Anthropic (47% less), GPT-5.3 Codex on OpenAI (68% less) and Command A Plus on AIHubMix (71% less). Savings are on a mix of 3 input tokens to 1 output token.
What is J2 Ultra's API model id?
Use j2-ultra with AI21 Studio. Other hosts use their own ids: ai21.j2-ultra-v1 on Amazon Bedrock.
Sources
Raw catalog entry for j2-ultra on AI21 Studio
{
"input_cost_per_token": 0.000015,
"litellm_provider": "ai21",
"max_input_tokens": 8192,
"max_output_tokens": 8192,
"mode": "completion",
"output_cost_per_token": 0.000015
}