FunSearch TPS calculator

Open weights Google DeepMind 15B parameters December 2023

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

306 cards that can run it

818 cards we hold specifications for

Smallest card that fits

P102-101

10 GB · IQ4_XS · 18.9 tok/s

Fastest card

B200

226 tok/s · 180 GB

Which GPUs can run FunSearch?

Set the inputs, read the answer

Context length tokens

A longer conversation needs more memory, which can push this model off smaller cards.

Minimum quality

Hides cards that would only fit the model by compressing it below this point.

Search cards

Card memory

Also show cards that cannot run it

306 cards match

Calculating

					Needs	Quantisation	Fit
226 tok/s 136–361 · low confidence	B200 NVIDIA	180 GB	8,000 GB/s	Jan 2024	16.8 GB	Q8_0	Comfortable
226 tok/s 136–361 · low confidence	B300 NVIDIA	288 GB	8,000 GB/s	Sep 2025	16.8 GB	Q8_0	Comfortable
180 tok/s 108–289 · low confidence	Radeon Instinct MI350X AMD	288 GB	8,190 GB/s	Jan 2025	16.8 GB	Q8_0	Comfortable
180 tok/s 108–289 · low confidence	Radeon Instinct MI355X AMD	288 GB	8,190 GB/s	Jan 2025	16.8 GB	Q8_0	Comfortable
144 tok/s 87–231 · low confidence	Radeon Instinct MI300 AMD	128 GB	6,550 GB/s	Jan 2023	16.8 GB	Q8_0	Comfortable
138 tok/s 83–221 · low confidence	H200 NVL NVIDIA	141 GB	4,890 GB/s	Nov 2024	16.8 GB	Q8_0	Comfortable
138 tok/s 83–221 · low confidence	H200 SXM 141 GB NVIDIA	141 GB	4,890 GB/s	Nov 2024	16.8 GB	Q8_0	Comfortable
132 tok/s 79–211 · low confidence	Radeon Instinct MI325X AMD	256 GB	6,000 GB/s	Oct 2024	16.8 GB	Q8_0	Comfortable
117 tok/s 70–188 · low confidence	Radeon Instinct MI300A AMD	128 GB	5,325 GB/s	Dec 2023	16.8 GB	Q8_0	Comfortable
117 tok/s 70–188 · low confidence	Radeon Instinct MI300X AMD	192 GB	5,325 GB/s	Dec 2023	16.8 GB	Q8_0	Comfortable
117 tok/s 70–188 · low confidence	Radeon Instinct MI308X AMD	192 GB	5,325 GB/s	Dec 2023	16.8 GB	Q8_0	Comfortable
111 tok/s 67–178 · low confidence	H100 NVL 94 GB NVIDIA	94 GB	3,940 GB/s	Mar 2023	16.8 GB	Q8_0	Comfortable
108 tok/s 65–173 · low confidence	CMP 170HX 10 GB NVIDIA	10 GB	1,560 GB/s	Sep 2021	8.9 GB	IQ4_XS	Tight
94.9 tok/s 57–152 · low confidence	H100 PCIe 96 GB NVIDIA	96 GB	3,360 GB/s	Mar 2023	16.8 GB	Q8_0	Comfortable
94.9 tok/s 57–152 · low confidence	H100 SXM5 80 GB NVIDIA	80 GB	3,360 GB/s	Oct 2022	16.8 GB	Q8_0	Comfortable
94.9 tok/s 57–152 · low confidence	H100 SXM5 94 GB NVIDIA	94 GB	3,360 GB/s	Mar 2023	16.8 GB	Q8_0	Comfortable
94.9 tok/s 57–152 · low confidence	H100 SXM5 96 GB NVIDIA	96 GB	3,360 GB/s	Mar 2023	16.8 GB	Q8_0	Comfortable
94.9 tok/s 57–152 · low confidence	H800 SXM5 NVIDIA	80 GB	3,360 GB/s	Mar 2023	16.8 GB	Q8_0	Comfortable
72.2 tok/s 43–116 · low confidence	Radeon Instinct MI250 AMD	128 GB	3,280 GB/s	Nov 2021	16.8 GB	Q8_0	Comfortable
72.2 tok/s 43–116 · low confidence	Radeon Instinct MI250X AMD	128 GB	3,280 GB/s	Nov 2021	16.8 GB	Q8_0	Comfortable
60.2 tok/s 36–96 · low confidence	Data Center GPU Max 1550 Intel	128 GB	3,280 GB/s	Jan 2023	16.8 GB	Q8_0	Comfortable
59.5 tok/s 36–95 · low confidence	GeForce RTX 3080 12 GB NVIDIA	12 GB	912 GB/s	Jan 2022	9.8 GB	Q4_K_M	Tight
59.5 tok/s 36–95 · low confidence	GeForce RTX 3080 Ti NVIDIA	12 GB	912 GB/s	May 2021	9.8 GB	Q4_K_M	Tight
58.9 tok/s 35–94 · low confidence	Data Center GPU Max Subsystem Intel	128 GB	3,210 GB/s	Jan 2023	16.8 GB	Q8_0	Comfortable
57.6 tok/s 35–92 · low confidence	A100 SXM4 80 GB NVIDIA	80 GB	2,040 GB/s	Nov 2020	16.8 GB	Q8_0	Comfortable

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: Google DeepMind
Organisation type: Industry
Country: United States of America
Published: 14 December 2023
Authors: Bernardino Romera-Paredes, Mohammadamin Barekatain, Alexander Novikov, Matej Balog, M. Pawan Kumar, Emilien Dupont, Francisco J. R. Ruiz, Jordan S. Ellenberg, Pengming Wang, Omar Fawzi, Pushmeet Kohli, Alhussein Fawzi

What it does

The problem areas the model was built for. A model can carry several of each.

Domain: Language, Search
Task: Code generation
Base model: PaLM 2

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: 15B
Training data: tokens

Training compute

The arithmetic performed to train the model, measured in floating-point operations. It is a measure of what the training run cost, not of how fast the finished model answers you.

Training compute: 3.9 × 10²³ FLOP
How it was established: Hardware

The training run

What it physically took to train: which chips, how many, for how long, and what that drew from the wall.

Wall-clock time: 48 hours

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 (unrestricted)
Training code: Unreleased

How it is classified

Labels the source dataset applies when tracking notable models, and how confident it is in the entry.

Likely above 10²³ FLOP: Yes
Why it is tracked: SOTA improvement,Historical significance
Record confidence: Speculative
Citations: 593

Sources

Where this record came from and when it was last checked.

Reference: Mathematical discoveries from program search with large language models
Last updated: 1 January 2026

The extremes

The ten fastest GPUs that run FunSearch

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 FunSearch

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

Hardware requirements in practice

Minimum card

P102-101

Memory needed

8.9 GB

Fastest

226 tok/s

FunSearch is small enough at 15B parameters that hardware is rarely the obstacle — 306 of the cards we track can run it, including cards several years old.

The entry point is the P102-101: 10 GB of memory, IQ4_XS compression, roughly 18.9 tokens per second.

At the other end, a B200 generates roughly 226 tokens per second on it, on the strength of 8,000 GB/s of memory bandwidth.

Background

FunSearch was published by Google DeepMind, in United States of America, in December 2023. It comes out of industry.

It works in Language, Search, and is recorded as doing code generation.

It is derived from PaLM 2 rather than trained from scratch, which is the usual way a specialised model is produced.

The weights are published, so it can be downloaded and run on your own hardware indefinitely, offline, with no account attached.

Reading the throughput figures

The median result is around 21.1 tokens per second; 266 cards produce text faster than most people read it.

It is a dense model, so every parameter is read for every token produced. That makes speed track memory bandwidth almost exactly — a card with twice the bandwidth generates roughly twice as fast.

Memory here is estimated from size rather than computed from the architecture, which is not recorded for this model — the numbers are indicative rather than exact.

Training and provenance

The training run consumed about 3.9 × 10²³ FLOP. That figure describes the cost of creating it and has no bearing on how quickly it generates text.

The reason it appears in this catalogue at all is sOTA improvement,Historical significance.

Step by step

How to choose a GPU for FunSearch

The table above has already assessed every card we hold specifications for against this model. Getting to your answer takes six steps.

01

Check what it needs before anything else

Every card here has been checked against FunSearch — around 8.9 GB at IQ4_XS. Capacity is the gate — a card either holds it or it does not.
02

Decide how long your conversations run

Longer conversations cost memory on top of what the weights need. Move the slider to your real working length before trusting any row for FunSearch.
03

Set a quality floor

The quantisation column varies by card, because a bigger card holds a more accurate copy of FunSearch — IQ4_XS on the smallest card that fits. Set a floor to hold the comparison at one level.
04

Compare tokens per second, not specifications

The speed ordering for FunSearch is effectively an ordering by memory bandwidth, which is why the B200 tops it at 226 tok/s.
05

Read the fit column last

The fit column separates cards that just manage FunSearch from those with room to spare. Buy for the second if the context might grow.
06

See what else that card runs

Each card page repeats this sweep for every model we hold. It answers what else the hardware is good for, beyond FunSearch.

Answers

FunSearch — common questions

How much compute was used to train FunSearch?

Around 3.9 × 10²³ FLOP. That measures what producing the model cost and says nothing about how quickly it answers once trained — inference speed comes from memory bandwidth, not from the training budget.

Can I run FunSearch if it does not fit in my GPU?

Only by offloading, which is usually a false economy: the part in system memory drags the whole thing down — the nearest miss we calculate is short by 2.6 GB. Our figures for FunSearch assume it is fully resident.

Would two GPUs run FunSearch faster?

A second card roughly doubles the memory available but not the generation rate. With 306 cards already able to run FunSearch alone, the case for pairing is weak.

Why does the quantisation differ between cards for FunSearch?

Because capacity varies, so does how hard FunSearch has to be squeezed — 4 distinct levels appear in the table above. Set a minimum quality to compare at one.

How accurate are these FunSearch speed estimates?

Every figure is derived from memory bandwidth and model size, not benchmarked. That is why each is published as a range such as 136–361 tok/s on the B200 rather than a single number.

What GPU do I need to run FunSearch?

The smallest card in our catalogue that holds FunSearch is the P102-101, with 10 GB of memory. It runs the model at IQ4_XS using about 8.9 GB, and produces roughly 18.9 tokens per second. 306 cards in total can run it.

How fast is FunSearch on a GPU?

It depends on the card. The quickest we calculate is a B200 at about 226 tokens per second; the slowest that still runs it manages considerably less. Reading speed is around ten tokens per second, and 266 of the cards that can run FunSearch clear that.

How much VRAM does FunSearch need?

About 8.9 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.

Can I run FunSearch on a 12 GB GPU?

Yes. A GeForce RTX 3080 Ti with 12 GB runs it at Q4_K_M, using about 9.8 GB and generating roughly 59.5 tokens per second — a tight fit.

Can I run FunSearch on a 16 GB GPU?

Yes. A Tesla V100 SXM2 16 GB with 16 GB runs it at Q6_K, using about 13.3 GB and generating roughly 46.4 tokens per second — a tight fit.

Can I run FunSearch on a 24 GB GPU?

Yes. A GeForce RTX 5090 D V2 with 24 GB runs it at Q8_0, using about 16.8 GB and generating roughly 37.8 tokens per second — a comfortable fit.

Is FunSearch open source?

Its weights are published, so FunSearch 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 FunSearch have?

FunSearch has 15B parameters. From the section called "Pretrained LLM": "We use Codey, an LLM built on top of the PaLM2 model family... Because FunSearch relies on sampling from an LLM extensively, an important performance-defining tradeoff is between the quality of the samples and the inference speed of the LLM. In practice, we have chosen to work with a fast-inference model (rather than slower-inference, higher-quality)" Unclear which PaLM2 model was used (of Gecko, Otter, Bison, and Unicorn); above quote indicates it was perhaps Otter or Bison, but not Unicorn. Exact parameter counts are not publicly disclosed for any of these models. In comparisons where FunSearch uses StarCoder-15B, Codey is an improvement but not obviously of an entirely different model class. I report the 15B parameters from StarCoder-15B, used as an open-source comparison. 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 FunSearch?

FunSearch was published by Google DeepMind, based in United States of America, categorised as industry.

When was FunSearch released?

FunSearch was published in December 2023. Capability per parameter has improved considerably since, so a newer model of the same size is often the better use of the same hardware.

What is FunSearch used for?

FunSearch works in Language, Search, and is recorded as handling code generation. A model can carry several of each, so these are the areas it was built for rather than a limit on what it will attempt.

Where can I download FunSearch?

The weights for FunSearch are published, though we do not hold a repository link for it. This site calculates hardware requirements rather than hosting model files.

Source

Original publication

Record last updated 1 January 2026

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.

All AI models