Share your thoughts, 1 month free Claude Pro on usSee more

Chart-to-code generation on ChartMimic Chart2Python (test)

48.5Error Rate

InternVL3.5-2B

Updated 1mo ago

Evaluation Results

Method	Links
InternVL3.5-2B 2026.04		48.5	65.6	32.5	31.3
DeepSeek-VL2-3B 2026.04		50.2	69.4	35	35
ChartMoE-8B 2026.04		52.7	56.9	22.9	25.3
Qwen3-VL-2B 2026.04		56.3	68.4	39.7	39.2
ChartLlama-13B 2026.04		57.5	44.9	28.1	24.8
LLaVA-v1.6-7B 2026.04		62	55.9	25.7	25.9
InternVL3.5-4B 2026.04		62.8	69.5	44.7	43
ChartLlama-13B 2026.04		65.3	46	14.8	16.2
Phi-3.5-vision-4B 2026.04		66.7	44.1	41	38.6
Phi-3.5-vision-4B 2026.04		68.8	53.3	56.1	34.2
ChartMoE-8B 2026.04		69.5	64.2	40.2	35.4
InternVL3.5-2B 2026.04		69.8	76.1	53.2	53.1
LLaVA-v1.6-7B 2026.04		71	65.7	38	39.5
InternVL3.5-8B 2026.04		71.2	71	52.1	48.9
DeepSeek-VL2-3B 2026.04		72	80.2	58.7	58
Qwen3-VL-4B 2026.04		72.5	71.9	58.4	55.2
Qwen3-VL-2B 2026.04		74	78	59.6	61
InternVL3.5-4B 2026.04		77.9	78.4	63.4	63
Qwen3-VL-8B 2026.04		78.7	72.5	65.2	61.9
InternVL3.5-8B 2026.04		82.5	79.6	67.5	67
CharLuMA-1.3B 2026.04		83	71.8	64.8	62.5
GPT-5-mini 2026.04		85.2	89	80	67.5
GPT-5-mini 2026.04		86.8	86.9	78.2	71.5
Qwen3-VL-4B 2026.04		87.6	83.2	77.2	76.1
Claude-Haiku-3.5 2026.04		88	76.2	73.5	65.7
GPT-4o-mini 2026.04		89	77.5	74.8	70.2
Qwen3-VL-8B 2026.04		91.1	83.7	80.8	80.6
Claude-Haiku-3.5 2026.04		91.3	81.6	76.7	68.8
ChartCoder-7B 2026.04		91.4	69.2	74	72.5
CharLuMA-6.7B 2026.04		92.3	77.4	75.1	73.3
GPT-4o 2026.04		93.2	83.5	83.5	79
CharLuMA-1.3B 2026.04		94.4	86.5	78.4	76.9
GPT-4o-mini 2026.04		94.8	81.2	79.8	74.5
Claude-Sonnet-4 2026.04		96.2	83.3	86.4	81.5
ChartCoder-7B 2026.04		96.2	48.1	86.4	56.1
CharLuMA-6.7B 2026.04		98	88.7	88.1	83.5
Claude-Sonnet-4 2026.04		98.3	86.8	88	81.4
GPT-4o 2026.04		98.5	85	87.4	80.9