Jamba 1.6 Large TPS calculator
Each card below is assessed against this model at the context length and minimum quality you choose. Speed is an estimate for a single request, calculated from the card's memory bandwidth and the size of the model once compressed.
Calculated for this model
818 cards we hold specifications for
Smallest card that fits
Radeon Instinct MI325X
256 GB · IQ4_XS · 51.8 tok/s
Fastest card
B300
83.2 tok/s · 288 GB
Which GPUs can run Jamba 1.6 Large?
Set the inputs, read the answer
A longer conversation needs more memory, which can push this model off smaller cards.
Hides cards that would only fit the model by compressing it below this point.
4 cards match
Calculating| Needs | Quantisation | Fit | |||||
|---|---|---|---|---|---|---|---|
|
83.2
tok/s
50–133 · low confidence |
B300 NVIDIA | 288 GB | 8,000 GB/s | Sep 2025 | 234.4 GB | Q4_K_M | Tight |
|
66.5
tok/s
40–106 · low confidence |
Radeon Instinct MI350X AMD | 288 GB | 8,190 GB/s | Jan 2025 | 234.4 GB | Q4_K_M | Tight |
|
66.5
tok/s
40–106 · low confidence |
Radeon Instinct MI355X AMD | 288 GB | 8,190 GB/s | Jan 2025 | 234.4 GB | Q4_K_M | Tight |
|
51.8
tok/s
31–83 · low confidence |
Radeon Instinct MI325X AMD | 256 GB | 6,000 GB/s | Oct 2024 | 211.3 GB | IQ4_XS | Tight |
Speeds are estimates for a single request — one conversation at a time — calculated from memory bandwidth, model size and quantisation. Real throughput varies with the inference runtime and its version. Figures published by hardware vendors measure many simultaneous requests and are much higher.
On record
Full specification
Everything on record for this model. Most of it describes how it was trained rather than how it runs — useful context for judging how much work went into it, and how it compares with models built at a different scale.
Origin
Who built this model, where, and when it was published.
- Organisation
- AI21 Labs
- Organisation type
- Industry
- Country
- Israel
- Published
- 6 March 2025
What it does
The problem areas the model was built for. A model can carry several of each.
- Domain
- Language
- Task
- Language modeling/generation, Question answering, Code generation, Retrieval-augmented generation, Chat, Quantitative reasoning, Table tasks, Translation
Size
How large the model is and how much data it was trained on. Parameters are the figure that decides whether it fits on a given graphics card.
- Parameters
- 398B
- Training data
- tokens
94B active/398B total
Availability
Whether you can obtain the model and run it on your own hardware, which is what decides if any of the graphics-card figures on this page apply.
- Weights
- Open — downloadable
- Model access
- Open weights (restricted use)
- Training code
- Unreleased
- Hugging Face
- ai21labs
Jamba Open Model License Agreement ($50M in annual revenue cap for commercial use): https://huggingface.co/ai21labs/AI21-Jamba-Large-1.6
How it is classified
Labels the source dataset applies when tracking notable models, and how confident it is in the entry.
- Record confidence
- Confident
Sources
Where this record came from and when it was last checked.
- Reference
- AI21’s Jamba 1.6: The Best Open Model for Private Enterprise Deployment
- Last updated
- 28 November 2025
The extremes
The ten fastest GPUs that run Jamba 1.6 Large
Ranked by estimated tokens per second, newest card first where speeds tie. Because generation is bound by memory bandwidth, this ordering follows bandwidth rather than any gaming benchmark.
The smallest GPUs that still run Jamba 1.6 Large
The cheapest route in, by memory capacity. A tight fit runs the model but leaves nothing spare for a longer conversation.
What the numbers mean
What it takes to run this model
Minimum card
Radeon Instinct MI325X
Memory needed
211.3 GB
Fastest
83.2 tok/s
At 398B parameters, Jamba 1.6 Large is beyond what any single graphics card holds. Running it means either splitting it across several cards or renting hardware built for the job — 4 of the cards we track can hold it on their own, and all of them are datacentre parts.
The smallest card that holds it is the Radeon Instinct MI325X with 256 GB, running it at IQ4_XS and producing around 51.8 tokens per second.
At the other end, a B300 generates roughly 83.2 tokens per second on it, on the strength of 8,000 GB/s of memory bandwidth.
About this model
Jamba 1.6 Large was published by AI21 Labs, in Israel, in March 2025. The organisation is categorised as industry.
It works in Language, and is recorded as doing language modeling/generation, Question answering, Code generation, Retrieval-augmented generation, Chat, Quantitative reasoning, Table tasks, Translation.
The weights being open is what puts this page in the calculator rather than only in the catalogue: it is a model you can actually hold. It is published under the ai21labs organisation on Hugging Face.
How fast it runs, and why
The median result is around 66.5 tokens per second; 4 cards produce text faster than most people read it.
This is a mixture-of-experts model, which routes each token through only part of itself. It therefore generates far faster than its total size suggests — while still needing every parameter resident in memory, so it is quick without being cheap to hold.
Its internal architecture is not on file, so memory is approximated from the parameter count and marked accordingly. Expect the real figure to differ, more so at long context.
Step by step
How to choose a GPU for Jamba 1.6 Large
The table above has already assessed every card we hold specifications for against this model. Getting to your answer takes six steps.
-
01
Start from the memory column
The table lists every card that can hold Jamba 1.6 Large — around 211.3 GB at IQ4_XS. That figure, not the card's headline performance, is what decides whether it runs.
-
02
Match the context to your actual use
The conversation occupies memory too, and grows as it goes. Set the slider to the length you expect: at long context Jamba 1.6 Large can slip off a card that handles short questions easily.
-
03
Set a quality floor
The quantisation column varies by card, because a bigger card holds a more accurate copy of Jamba 1.6 Large — IQ4_XS on the smallest card that fits. Set a floor to hold the comparison at one level.
-
04
Rank by throughput rather than spec sheet
Sort by speed to see how cards rank for Jamba 1.6 Large. It will not match a gaming ordering — generation is bound by memory bandwidth, which is why the B300 tops it at 83.2 tok/s.
-
05
Read the fit column last
Tight means Jamba 1.6 Large loads and works, with no room to raise the context later. Comfortable means you can. The difference matters more than a few tokens per second.
-
06
Check the card from the other side
Each card page repeats this sweep for every model we hold. It answers what else the hardware is good for, beyond Jamba 1.6 Large.
Answers
Jamba 1.6 Large — common questions
What is Jamba 1.6 Large used for?
Jamba 1.6 Large works in Language, and is recorded as handling language modeling/generation, Question answering, Code generation, Retrieval-augmented generation, Chat, Quantitative reasoning, Table tasks, Translation. Models frequently carry more than one of each, and the tags describe purpose rather than capability limits.
Where can I download Jamba 1.6 Large?
Its weights are published under the ai21labs organisation on Hugging Face. We do not host model files — this site calculates what hardware is needed to run them.
Can I run Jamba 1.6 Large if it does not fit in my GPU?
It can be split between the card and system memory, but Jamba 1.6 Large generates painfully slowly that way — the nearest miss we calculate is short by 61.6 GB. Nothing on this page assumes offloading.
Would two GPUs run Jamba 1.6 Large faster?
A second card roughly doubles the memory available but not the generation rate. With 4 cards already able to run Jamba 1.6 Large alone, the case for pairing is weak.
Why does the quantisation differ between cards for Jamba 1.6 Large?
Because capacity varies, so does how hard Jamba 1.6 Large has to be squeezed — 2 distinct levels appear in the table above. Set a minimum quality to compare at one.
How accurate are these Jamba 1.6 Large speed estimates?
They are calculated from specifications rather than measured, and each carries a range — 50–133 tok/s on the B300, for instance. The same model and card vary by thirty to fifty per cent depending on the inference software and its version.
What GPU do I need to run Jamba 1.6 Large?
The smallest card in our catalogue that holds Jamba 1.6 Large is the Radeon Instinct MI325X, with 256 GB of memory. It runs the model at IQ4_XS using about 211.3 GB, and produces roughly 51.8 tokens per second. 4 cards in total can run it.
How fast is Jamba 1.6 Large on a GPU?
It depends on the card. The quickest we calculate is a B300 at about 83.2 tokens per second; the slowest that still runs it manages considerably less. Reading speed is around ten tokens per second, and 4 of the cards that can run Jamba 1.6 Large clear that.
How much VRAM does Jamba 1.6 Large need?
About 211.3 GB at IQ4_XS compression, which is what the smallest card that runs it uses. Less compression needs more: the figures in the memory column above are recalculated for each card, because each one holds the least-compressed version it can.
Is Jamba 1.6 Large open source?
Its weights are published, so Jamba 1.6 Large can be downloaded and run on your own hardware. Note that open weights is not the same as open source in the full sense — it says nothing about the training data, the training code, or the commercial terms attached.
How many parameters does Jamba 1.6 Large have?
Jamba 1.6 Large has 398B parameters. 94B active/398B total. That figure is the total, and it is what decides how much memory the model needs — roughly half a gigabyte per billion at the compression most people use.
Who created Jamba 1.6 Large?
Jamba 1.6 Large was published by AI21 Labs, based in Israel, categorised as industry.
When was Jamba 1.6 Large released?
Jamba 1.6 Large was published in March 2025.
The other direction
Looking at it from the other side?
This page starts from the model. If you already own a card and want to know everything it will run, start from the hardware instead.