← comparisons · head to head
Muse Spark Open vs Kimi K3
Kimi K3 scores higher than Muse Spark Open on all 4 public leaderboards that rate both. On the ArtificialWatch Index, which takes the median of each model's percentile across the boards that score it, Muse Spark Open is at 16.7 and Kimi K3 at 37 (out of 100).
Scores as published by each board; table composed November 13, 2026.
Get the next launch alert, free →
Leaderboard by leaderboard
| Board | Muse Spark Open | Kimi K3 | Higher |
|---|---|---|---|
| LMArena Elo (community votes) | 1,526.3 | 1,657.8 | Kimi K3 |
| LiveBench global average (0-100) | 48.5 | 65 | Kimi K3 |
| Vals AI Vals Index accuracy % · weighted finance + coding tasks | 48.6 | 53 | Kimi K3 |
| Epoch AI Epoch Capabilities Index (ECI) · many benchmarks on one scale | 155.4 | 158.6 | Kimi K3 |
Raw scores are each board's own scale, so compare within a row, not across rows. Full table: the bench index.
For coding
On the ArtificialWatch coding index, Muse Spark Open is at 21.6 (#43 of 48) and Kimi K3 at 64 (#17). Muse Spark Open scores higher on 1 of the 2 coding boards that rate both and Kimi K3 on 1.
| Coding board | Muse Spark Open | Kimi K3 |
|---|---|---|
| LiveBench · Coding | 55.1 | 53.1 |
| Vals Vibe Code Bench | 76.2 | 85 |
Muse Spark Open at a glance
- Developer
- Meta
- Status
- Live since October 15, 2026
Live — Oct 15 · Meta's open-weights Spark: 816B total / 42B active mixture-of-experts with a 1.8B perception encoder, 1M context, text, image, video and audio in, text out · reasoning effort low / medium / high / xhigh, default xhigh · weights on Hugging Face in BF16 and NVFP4, gated, under Meta's Muse Model License · Meta says NVFP4 needs at least 700GB of VRAM and BF16 2TB, and that it is tuned to run on two M5 Ultra Mac Studios with MLX · OpenRouter: $1.25 / $4.25 per 1M through BaseTen (FP4), the same price as Muse Spark 1.3 · Meta's own table, against open-weight models: Artificial Analysis Intelligence Index 39, behind GLM 5.3 (45), Kimi K3 (44) and Qwen3.8 (40); ahead of all of them on JobBench (61.8) and MCP-Atlas (89.2); behind on Terminal-Bench 2.1 (76.4 vs Kimi K3's 85.0) · Artificial Analysis has no page of its own for it yet (Oct 15).
Meta releases Muse Spark Open: 816B open weights at $1.25 / $4.25, and its own table puts it behind Kimi K3 and GLM 5.3
2026-10-15 · Meta has released Muse Spark Open, an open-weights version of its Spark model: 816 billion parameters in total, 42 billion active per token, with a 1M-token context. Meta's model card on Hugging Face dates the release October 15.
Kimi K3 at a glance
- Developer
- Moonshot
- Status
- Live since July 15, 2026
- API price
- $3 input / $15 output per 1M tokens · Moonshot AI, kimi-k3
- Context window
- 1.05M tokens
Live — dropped Jul 15 · 2.8T · open weights landed Jul 27 (1.56 TB, MXFP4).
Kimi K3, day one: impressive, expensive, and a little slow
2026-07-16 · Kimi K3 has been live for a day, and the picture forming is more interesting than the launch-night hype. Here's everything we're seeing so far.
Muse Spark Open vs Kimi K3: quick answers
Which is better, Muse Spark Open or Kimi K3?
It depends on what you measure. Kimi K3 scores higher than Muse Spark Open on all 4 public leaderboards that rate both. Kimi K3's widest lead is on LMArena. On the ArtificialWatch Index, which takes the median of each model's percentile across the boards that score it, Muse Spark Open is at 16.7 and Kimi K3 at 37 (out of 100).
Which is newer, Muse Spark Open or Kimi K3?
Muse Spark Open. It went live on October 15, 2026, 92 days after Kimi K3 (July 15, 2026).
How do Muse Spark Open and Kimi K3 compare for coding?
On the ArtificialWatch coding index, Muse Spark Open is at 21.6 (#43 of 48) and Kimi K3 at 64 (#17). Muse Spark Open scores higher on 1 of the 2 coding boards that rate both and Kimi K3 on 1.
Who makes Muse Spark Open and Kimi K3?
Muse Spark Open is from Meta; Kimi K3 is from Moonshot.
Where do these numbers come from?
Each score is the leaderboard's own published figure, read by ArtificialWatch (table composed November 13, 2026). Boards measure different things, so raw scores are only comparable within a row.
More comparisons
Hear when the next one drops. Free.
Free: a Chrome push and an email about 15 minutes after a new model answers on a public API. Members get a text or a phone call the moment a launch is confirmed.
- Free forever
- No card
- Unsubscribe in one click
Also: the bench index · API pricing · the wire