whatmodel.sh / Palmyra X4 (palmyra-x-004)
Palmyra X4 (palmyra-x-004)
↗ share
+ contribute
Palmyra X4 (palmyra-x-004) is a closed-weights language model from Writer, released October 2024. Licence: proprietary.
Catalogue identity notes
Comparison-only. Writer's Palmyra X4, evaluated under the model string 'palmyra-x-004' by BFCL (entry 'palmyra-x-004 (FC)', Organization 'Writer', Model Link writer.com/engineering/actions-with-palmyra-x-004/) and Galileo (row 'palmyra-x-004', Vendor 'Writer'). That linked Writer post is titled 'Introducing actions with Palmyra X4' (Sam Julien, October 9, 2024), tying the string to Palmyra X4. Writer's docs (dev.writer.com/home/models.md, observed 2026-10-09) list Palmyra X4 with API model ID `palmyra-x4` (Bedrock writer.palmyra-x4-v1:0, 128k context, $2.50/$10.00 per 1M tokens), deprecated 2026-12-14; no Writer page states that `palmyra-x4` serves the same checkpoint as the 'palmyra-x-004' id the evaluators called, so no run route or price is attached. Proposed so its BFCL v3, BFCL V4 and Galileo v1 rows (three multi-model runs) count. New provider string 'Writer' (no logo yet).
Model information sources
$ how it compares
The LLMs ranked either side of Palmyra X4 (palmyra-x-004), out of 828 with a score.
| # | model | score | where |
|---|---|---|---|
| 380 |  Granite 4.0 H Small (32B-A9B) | 52±1 | api $0.11/M +local |
| 381 |  Llama 3.1 Nemotron 70B Instruct | 52±1 | local |
| 382 | Palmyra X4 (palmyra-x-004) | 52±4 | comparison only |
| 383 | Nemotron-4 340B Instruct | 52±2 | comparison only |
| 384 |  Aya Expanse 32B | 52±2 | api $0.75/M +local |
$ what it's good at
agent tasks▁▁▂▆█▆▅▄▄▃▂▁ #258 of 340 · 94% of its score · 3 tests
Beats o3-mini, Nova 2 Lite, Ministral 3 14B Reasoning 2512 here, though they rank higher overall. Trails SmolLM3 3B, Jamba 1.5 Mini, LFM2.5 1.2B Thinking here, though they rank lower overall.
| test | result | vs the others in its run | counts |
|---|---|---|---|
| BFCL V4 (December 2025 archive) | 27.9% ✎ | ▁▁▁▂██▅▅▆▂▂▂ beats 23/77 | 50% |
| BFCL v3 | 56.8% ✎ | ▁▁▁▁▁▃▅█▅▄▁▁ beats 65/114 | 25% |
| Galileo Agent Leaderboard v1 (tool selection quality) | 88.6% ✎ | ▃▃▃▃▆▅▆▅█▂▅▂ beats 19/31 | 20% |
$ where to run it
No verified API or download recorded yet · add one
$ how the score was worked out · Berkeley Function Calling Lea… 75%, Galileo Agent Leaderboard (v1… 20%
Each result implies a score on its own, after adjusting for how hard its test is and who ran it. The score is their weighted average, with a small pull toward the middle for thinly tested models. Each test's weight is its counts column above; here it's added up by who ran the tests. How it works
| source | results | share of the score |
|---|---|---|
 Berkeley Function Calling Leaderboard result archive | 2 | 75% |
| Galileo Agent Leaderboard (v1 and v2 results) | 1 | 20% |
| prior · pull toward the middle | 6% | |
| weighted average | = 51.8 |
$ improve this page
Use a short GitHub form or edit the data through a pull request. Every number needs a source; here's which ones we can use.
data updated · sources and licences · json