harn-vm 0.8.144

Async bytecode virtual machine for the Harn programming language
Documentation 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188

# Baseten Model APIs. Live /v1/models verified 2026-06-23 with
# BASETEN_API_KEY: ids, served context windows, pricing, features, modalities,
# and route quantization come from the provider-owned endpoint. Rows are keyed
# with `baseten/<wire-id>` so failover/eval tooling can compare the same
# weights across Baseten, Z.AI, Together, DeepInfra, OpenRouter, and NVIDIA.

[models."baseten/zai-org/GLM-5.2"]
name = "GLM 5.2 (Baseten)"
provider = "baseten"
wire_model = "zai-org/GLM-5.2"
logical_model = "glm-5.2"
equivalence_group = "glm-5.2"
served_variant = "baseten-model-api"
api_dialect = "openai_chat"
context_window = 262144
capabilities = ["tools", "streaming", "thinking", "prompt_caching"]
pricing = { input_per_mtok = 1.40, output_per_mtok = 4.40, cache_read_per_mtok = 0.26 }
architecture = { parameter_count_b = 744.0, active_parameter_count_b = 40.0, moe = true, quantization = "fp8", license = "MIT", source_url = "https://www.baseten.co/blog/how-we-built-the-worlds-fastest-api-for-glm-52/", last_verified = "2026-06-23" }
performance = { output_tokens_per_sec = 281.2, time_to_answer_s = 7.95, source = "artificial_analysis", source_url = "https://artificialanalysis.ai/models/glm-5-2/providers", last_verified = "2026-06-23", notes = "Artificial Analysis GLM-5.2 provider benchmark; Harn 2026-06-23 smoke probes confirmed non-thinking and thinking Baseten calls plus native-tool unreliability." }
tier = "frontier"
open_weight = true
strengths = ["speed", "coding", "agentic", "tool_use", "reasoning", "long_context"]
complementary_with = ["anthropic-claude", "openai-gpt", "google-gemini", "kimi", "deepseek"]

[models."baseten/moonshotai/Kimi-K2.7-Code"]
name = "Kimi K2.7 Code (Baseten)"
provider = "baseten"
wire_model = "moonshotai/Kimi-K2.7-Code"
logical_model = "moonshot-kimi-k2.7-code"
equivalence_group = "moonshot-kimi-k2.7-code"
served_variant = "baseten-model-api"
api_dialect = "openai_chat"
context_window = 262000
capabilities = ["tools", "vision", "streaming", "thinking", "prompt_caching"]
pricing = { input_per_mtok = 0.95, output_per_mtok = 4.00, cache_read_per_mtok = 0.16 }
architecture = { quantization = "fp4", source_url = "https://www.baseten.co/library/kimi-k27-code/", last_verified = "2026-06-23" }
tier = "frontier"
open_weight = true
strengths = ["coding", "agentic", "tool_use", "reasoning", "long_context", "vision"]
complementary_with = ["anthropic-claude", "openai-gpt", "google-gemini", "qwen", "deepseek"]

[models."baseten/deepseek-ai/DeepSeek-V4-Pro"]
name = "DeepSeek V4 Pro (Baseten)"
provider = "baseten"
wire_model = "deepseek-ai/DeepSeek-V4-Pro"
logical_model = "deepseek-v4-pro"
equivalence_group = "deepseek-v4-pro"
served_variant = "baseten-model-api"
api_dialect = "openai_chat"
context_window = 131000
capabilities = ["tools", "streaming", "thinking", "prompt_caching"]
pricing = { input_per_mtok = 1.74, output_per_mtok = 3.48, cache_read_per_mtok = 0.145 }
architecture = { parameter_count_b = 1600.0, active_parameter_count_b = 49.0, moe = true, quantization = "fp4", source_url = "https://www.baseten.co/library/deepseek-v4/", last_verified = "2026-06-23" }
tier = "frontier"
open_weight = true
strengths = ["reasoning", "coding", "agentic", "tool_use", "long_context"]
complementary_with = ["anthropic-claude", "openai-gpt", "google-gemini", "qwen", "kimi"]

[models."baseten/nvidia/NVIDIA-Nemotron-3-Ultra-550B-A55B"]
name = "Nemotron 3 Ultra 550B A55B (Baseten)"
provider = "baseten"
wire_model = "nvidia/NVIDIA-Nemotron-3-Ultra-550B-A55B"
logical_model = "nemotron-3-ultra-550b-a55b"
equivalence_group = "nemotron-3-ultra-550b-a55b"
served_variant = "baseten-model-api"
api_dialect = "openai_chat"
context_window = 202800
capabilities = ["tools", "streaming", "thinking", "prompt_caching"]
pricing = { input_per_mtok = 0.60, output_per_mtok = 2.40, cache_read_per_mtok = 0.12 }
architecture = { parameter_count_b = 550.0, active_parameter_count_b = 55.0, moe = true, quantization = "fp4", source_url = "https://www.baseten.co/library/nemotron-3-ultra/", last_verified = "2026-06-23" }
tier = "frontier"
open_weight = true
strengths = ["reasoning", "coding", "agentic", "tool_use", "long_context"]
complementary_with = ["anthropic-claude", "openai-gpt", "google-gemini", "qwen", "deepseek", "kimi"]

[models."baseten/nvidia/Nemotron-120B-A12B"]
name = "Nemotron 3 Super 120B A12B (Baseten)"
provider = "baseten"
wire_model = "nvidia/Nemotron-120B-A12B"
logical_model = "nemotron-3-super-120b-a12b"
equivalence_group = "nemotron-3-super-120b-a12b"
served_variant = "baseten-model-api"
api_dialect = "openai_chat"
context_window = 202800
capabilities = ["tools", "streaming", "thinking", "prompt_caching"]
pricing = { input_per_mtok = 0.30, output_per_mtok = 0.75, cache_read_per_mtok = 0.06 }
architecture = { parameter_count_b = 120.0, active_parameter_count_b = 12.0, moe = true, quantization = "fp4", source_url = "https://www.baseten.co/library/nemotron-3-super/", last_verified = "2026-06-23" }
tier = "frontier"
open_weight = true
strengths = ["speed", "cheap", "reasoning", "coding", "agentic", "tool_use", "long_context"]
complementary_with = ["anthropic-claude", "openai-gpt", "google-gemini", "qwen", "deepseek", "kimi"]

[models."baseten/openai/gpt-oss-120b"]
name = "GPT-OSS 120B (Baseten)"
provider = "baseten"
wire_model = "openai/gpt-oss-120b"
logical_model = "openai-gpt-oss-120b"
equivalence_group = "openai-gpt-oss-120b"
served_variant = "baseten-model-api"
api_dialect = "openai_chat"
context_window = 128072
capabilities = ["tools", "streaming", "thinking", "prompt_caching"]
pricing = { input_per_mtok = 0.10, output_per_mtok = 0.50, cache_read_per_mtok = 0.10 }
architecture = { parameter_count_b = 117.0, active_parameter_count_b = 5.1, moe = true, quantization = "fp4", license = "Apache-2.0", source_url = "https://www.baseten.co/library/gpt-oss-120b/", last_verified = "2026-06-23" }
tier = "frontier"
open_weight = true
strengths = ["cheap", "tool_use", "reasoning", "coding"]
complementary_with = ["anthropic-claude", "google-gemini", "qwen", "deepseek", "kimi"]

[models."baseten/zai-org/GLM-5.1"]
name = "GLM 5.1 (Baseten)"
provider = "baseten"
wire_model = "zai-org/GLM-5.1"
logical_model = "glm-5.1"
equivalence_group = "glm-5.1"
served_variant = "baseten-model-api"
api_dialect = "openai_chat"
context_window = 202800
capabilities = ["tools", "streaming", "prompt_caching"]
pricing = { input_per_mtok = 1.30, output_per_mtok = 4.30, cache_read_per_mtok = 0.26 }
architecture = { quantization = "fp4", source_url = "https://www.baseten.co/pricing/", last_verified = "2026-06-23" }
tier = "frontier"
open_weight = true
strengths = ["coding", "agentic", "tool_use", "long_context"]

[models."baseten/zai-org/GLM-5"]
name = "GLM 5 (Baseten)"
provider = "baseten"
wire_model = "zai-org/GLM-5"
logical_model = "glm-5"
equivalence_group = "glm-5"
served_variant = "baseten-model-api"
api_dialect = "openai_chat"
context_window = 202800
capabilities = ["tools", "streaming", "prompt_caching"]
pricing = { input_per_mtok = 0.95, output_per_mtok = 3.15, cache_read_per_mtok = 0.20 }
architecture = { quantization = "fp4", source_url = "https://www.baseten.co/pricing/", last_verified = "2026-06-23" }
tier = "frontier"
open_weight = true
strengths = ["coding", "agentic", "tool_use", "long_context"]

[models."baseten/zai-org/GLM-4.7"]
name = "GLM 4.7 (Baseten)"
provider = "baseten"
wire_model = "zai-org/GLM-4.7"
logical_model = "glm-4.7"
equivalence_group = "glm-4.7"
served_variant = "baseten-model-api"
api_dialect = "openai_chat"
context_window = 200000
capabilities = ["tools", "streaming", "prompt_caching"]
pricing = { input_per_mtok = 0.60, output_per_mtok = 2.20, cache_read_per_mtok = 0.12 }
architecture = { quantization = "fp4", source_url = "https://www.baseten.co/pricing/", last_verified = "2026-06-23" }
tier = "frontier"
open_weight = true
strengths = ["coding", "agentic", "tool_use", "long_context"]

[models."baseten/moonshotai/Kimi-K2.6"]
name = "Kimi K2.6 (Baseten)"
provider = "baseten"
wire_model = "moonshotai/Kimi-K2.6"
logical_model = "moonshot-kimi-k2.6"
equivalence_group = "moonshot-kimi-k2.6"
served_variant = "baseten-model-api"
api_dialect = "openai_chat"
context_window = 262000
capabilities = ["tools", "vision", "streaming", "thinking", "prompt_caching"]
pricing = { input_per_mtok = 0.95, output_per_mtok = 4.00, cache_read_per_mtok = 0.16 }
architecture = { quantization = "fp4", source_url = "https://www.baseten.co/pricing/", last_verified = "2026-06-23" }
tier = "frontier"
open_weight = true
strengths = ["coding", "agentic", "tool_use", "reasoning", "long_context", "vision"]

[models."baseten/moonshotai/Kimi-K2.5"]
name = "Kimi K2.5 (Baseten)"
provider = "baseten"
wire_model = "moonshotai/Kimi-K2.5"
logical_model = "moonshot-kimi-k2.5"
equivalence_group = "moonshot-kimi-k2.5"
served_variant = "baseten-model-api"
api_dialect = "openai_chat"
context_window = 262000
capabilities = ["tools", "vision", "streaming", "prompt_caching"]
pricing = { input_per_mtok = 0.60, output_per_mtok = 3.00, cache_read_per_mtok = 0.12 }
architecture = { quantization = "fp4", source_url = "https://www.baseten.co/pricing/", last_verified = "2026-06-23" }
tier = "frontier"
open_weight = true
strengths = ["coding", "agentic", "tool_use", "long_context", "vision"]