Kimi K3 is competitive with Fable; Kimi K3 + Fable is SoTA.

Announcing our Series D and $1B ARR

Log In

Get Started

Blog

Kimik3 Fable<br>Kimi K3 is competitive with Fable; Kimi K3 + Fable is SoTA.

PUBLISHED 7/21/2026

Table of Contents<br>How We Measured<br>K3 is a good model.<br>Two Models is Better than One<br>K3 can be up to 50x lower cost on Fireworks. 🫳🎤<br>Don't pick a model. Route.<br>K3 is cost optimized on all work types<br>Single Models Are Wasteful and No Longer SoTA

Table of Contents

Explore us in AI toolsChatGPTClaudeGrokPerplexityCoPilotGemini

Table of Contents<br>How We Measured<br>K3 is a good model.<br>Two Models is Better than One<br>K3 can be up to 50x lower cost on Fireworks. 🫳🎤<br>Don't pick a model. Route.<br>K3 is cost optimized on all work types<br>Single Models Are Wasteful and No Longer SoTA

Table of Contents

K3 is a frontier quality open model at a fraction of the cost. Even bigger is that it complements Fable predictably, which makes it possible to get the highest quality intelligence by routing tasks.

🧭 tl;dr: We ran Kimi K3 (open) against Fable 5 (closed) on ~1,000 agentic tasks finding:<br>We achieved 93% accuracy with routing between K3 and Fable.<br>Results were up to ~50X more cost effective than Fable alone on long agentic loops, and consistently lower cost across every use case.<br>How We Measured

We averaged benchmarks, each aimed at a different kind of work , and ran K3 and Fable 5 through the same harness. About 1,030 tasks in all, in real agent loops.<br>FamilyWhat it testsTasksSWEReal repo bug-fixes (SWE-bench style)460TerminalLong agentic ops: security, crypto, reverse-eng, sysadmin89AlgorithmicLeetCode / AtCoder-style problems100Multi-LanguageImplementation across six languages225LegalA legal-agent benchmark (lawyer-graded tasks)120<br>One quick definition before we get into the results. Oracle routing is a method for measuring the best theoretical performance by running the task through each model and then picking the cheapest correct option (the cost/performance ceiling). In a practical router, you don’t get to run your task against multiple models. The router makes a prediction of which model has the best cost and quality trade off, but ultimately it’s a guess.<br>In this study, oracle routing demonstrated K3 is selected for 72-96% of tasks. This suggests a near-perfect router might be achievable, by learning the difference between day-to-day tasks and the true long tail of frontier work. It will require an order of magnitude more routing data, and real world performance to say definitively.<br>K3 is a good model.

From a 10,000 foot view, it can be easy to look at both models and call the head-to-head a tie. For example, if you look at SWE, the headline benchmark, K3 gets 92.4% , Fable 92.6% . Across the five types of tasks we benchmarked on, the two models tend to stay within a few points of each other, with Fable pulling slightly ahead on its coding-language breadth (Multi-lang).<br>Fig 1 · Task solve rate by category. Near-identical on the overall average; good at different things underneath.It’s easy to stop there and say “they’re roughly even”. The news is that they have discretely better performance across different task types.<br>Two Models is Better than One

If you take a peek inside a single benchmark, there’s more to see than just a top-line accuracy number. Take SWE, where the two are dead even overall. If you split SWE by problem domain you can see where each model shines. K3 is sharpest on symbolic math and dev tooling; Fable wins on web & data visualization work. The same pattern runs through the multi-language set, where Fable's breadth carries Java, Python and C++, while K3 draws even on JavaScript and Rust.<br>Fig 2 · Point margin by SWE domain (K3 minus Fable). Tied on the benchmark as a whole, the two still specialize domain by domain.For long-horizon work at a terminal, driving a shell and prodding at systems across dozens of turns, K3 showed its true colors. It cleared a batch of tasks Fable never cracked: a 7z hash, FEAL cryptanalysis, leaked secrets, a live vulnerability, runaway async jobs.<br>Fig 3 · Of 89 terminal tasks, K3 has 11 solo wins to Fable's 7, and takes the security and crypto cluster outright.K3 can be up to 50x lower cost on Fireworks. 🫳🎤

While quality is a near-tie at a high level, price isn't close.<br>Fig 4 · Cost advantage per task. Two things behind it: token pricing, and the fact that which model runs longer depends on the task.So where's this huge price gap coming from? token pricing, prompt caching, and effort-per-task. On SWE for example, K3 works much harder than Fable: roughly 55 turns and 1.3M tokens a task versus 21 turns and 130K. On the long terminal tasks it's the other way around: Fable is the one that spirals, running up 64 turns and 1.5M tokens (sometimes straight into a timeout).<br>Fig 5 · Turns (linear) and tokens (log) per task. Neither model is leaner across the board; the extra work falls on SWE for K3 and on the terminal tasks for Fable. Note the y-axis for...